main.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pickle\n",
    "\n",
    "import tensorflow as tf\n",
    "from keras.callbacks import *\n",
    "\n",
    "from data_loader import *\n",
    "from model import *\n",
    "from my_metrics import *\n",
    "\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"0\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "訓練跟驗證集的原圖與標籤增生\n",
    "可以將trainGenerator函式中save_to_dir改到自己指定資料夾, 就能看到增生的圖像與標籤\n",
    "例如 save_to_dir=r\"data\\membrane\\train\\aug\"\n",
    "'''\n",
    "\n",
    "data_gen_args = dict(rotation_range=0.2,\n",
    "                     width_shift_range=0.05,\n",
    "                     height_shift_range=0.05,\n",
    "                     shear_range=0.05,\n",
    "                     zoom_range=0.05,\n",
    "                     horizontal_flip=True,\n",
    "                     fill_mode=\"nearest\")\n",
    "\n",
    "myGene = train_generator(16, \"data/membrane/train\", \"image\", \"label\", data_gen_args, save_to_dir=None)\n",
    "\n",
    "val_aug_dict = {\"rescale\": 1./255}\n",
    "\n",
    "valGene = train_generator(16, \"data/membrane/val\", \"val_image\", \"val_label\", val_aug_dict,\n",
    "                         image_color_mode=\"grayscale\", mask_color_mode=\"grayscale\",\n",
    "                         image_save_prefix=\"val_image\", mask_save_prefix=\"val_label\",\n",
    "                         save_to_dir=None, target_size=(256, 256), seed=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input_1 (InputLayer)           [(None, 256, 256, 1  0           []                               \n",
      "                                )]                                                                \n",
      "                                                                                                  \n",
      " batch_normalization (BatchNorm  (None, 256, 256, 1)  4          ['input_1[0][0]']                \n",
      " alization)                                                                                       \n",
      "                                                                                                  \n",
      " activation (Activation)        (None, 256, 256, 1)  0           ['batch_normalization[0][0]']    \n",
      "                                                                                                  \n",
      " conv2d (Conv2D)                (None, 256, 256, 16  160         ['activation[0][0]']             \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_1 (BatchNo  (None, 256, 256, 16  64         ['conv2d[0][0]']                 \n",
      " rmalization)                   )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_2 (Conv2D)              (None, 256, 256, 16  32          ['input_1[0][0]']                \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " activation_1 (Activation)      (None, 256, 256, 16  0           ['batch_normalization_1[0][0]']  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_2 (BatchNo  (None, 256, 256, 16  64         ['conv2d_2[0][0]']               \n",
      " rmalization)                   )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_1 (Conv2D)              (None, 256, 256, 16  2320        ['activation_1[0][0]']           \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " add (Add)                      (None, 256, 256, 16  0           ['batch_normalization_2[0][0]',  \n",
      "                                )                                 'conv2d_1[0][0]']               \n",
      "                                                                                                  \n",
      " max_pooling2d (MaxPooling2D)   (None, 128, 128, 16  0           ['add[0][0]']                    \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_3 (BatchNo  (None, 128, 128, 16  64         ['max_pooling2d[0][0]']          \n",
      " rmalization)                   )                                                                 \n",
      "                                                                                                  \n",
      " activation_2 (Activation)      (None, 128, 128, 16  0           ['batch_normalization_3[0][0]']  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_3 (Conv2D)              (None, 128, 128, 32  4640        ['activation_2[0][0]']           \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_4 (BatchNo  (None, 128, 128, 32  128        ['conv2d_3[0][0]']               \n",
      " rmalization)                   )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_5 (Conv2D)              (None, 128, 128, 32  544         ['max_pooling2d[0][0]']          \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " activation_3 (Activation)      (None, 128, 128, 32  0           ['batch_normalization_4[0][0]']  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_5 (BatchNo  (None, 128, 128, 32  128        ['conv2d_5[0][0]']               \n",
      " rmalization)                   )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_4 (Conv2D)              (None, 128, 128, 32  9248        ['activation_3[0][0]']           \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " add_1 (Add)                    (None, 128, 128, 32  0           ['batch_normalization_5[0][0]',  \n",
      "                                )                                 'conv2d_4[0][0]']               \n",
      "                                                                                                  \n",
      " max_pooling2d_1 (MaxPooling2D)  (None, 64, 64, 32)  0           ['add_1[0][0]']                  \n",
      "                                                                                                  \n",
      " batch_normalization_6 (BatchNo  (None, 64, 64, 32)  128         ['max_pooling2d_1[0][0]']        \n",
      " rmalization)                                                                                     \n",
      "                                                                                                  \n",
      " activation_4 (Activation)      (None, 64, 64, 32)   0           ['batch_normalization_6[0][0]']  \n",
      "                                                                                                  \n",
      " conv2d_6 (Conv2D)              (None, 64, 64, 64)   18496       ['activation_4[0][0]']           \n",
      "                                                                                                  \n",
      " batch_normalization_7 (BatchNo  (None, 64, 64, 64)  256         ['conv2d_6[0][0]']               \n",
      " rmalization)                                                                                     \n",
      "                                                                                                  \n",
      " conv2d_8 (Conv2D)              (None, 64, 64, 64)   2112        ['max_pooling2d_1[0][0]']        \n",
      "                                                                                                  \n",
      " activation_5 (Activation)      (None, 64, 64, 64)   0           ['batch_normalization_7[0][0]']  \n",
      "                                                                                                  \n",
      " batch_normalization_8 (BatchNo  (None, 64, 64, 64)  256         ['conv2d_8[0][0]']               \n",
      " rmalization)                                                                                     \n",
      "                                                                                                  \n",
      " conv2d_7 (Conv2D)              (None, 64, 64, 64)   36928       ['activation_5[0][0]']           \n",
      "                                                                                                  \n",
      " add_2 (Add)                    (None, 64, 64, 64)   0           ['batch_normalization_8[0][0]',  \n",
      "                                                                  'conv2d_7[0][0]']               \n",
      "                                                                                                  \n",
      " max_pooling2d_2 (MaxPooling2D)  (None, 32, 32, 64)  0           ['add_2[0][0]']                  \n",
      "                                                                                                  \n",
      " batch_normalization_9 (BatchNo  (None, 32, 32, 64)  256         ['max_pooling2d_2[0][0]']        \n",
      " rmalization)                                                                                     \n",
      "                                                                                                  \n",
      " activation_6 (Activation)      (None, 32, 32, 64)   0           ['batch_normalization_9[0][0]']  \n",
      "                                                                                                  \n",
      " conv2d_9 (Conv2D)              (None, 32, 32, 128)  73856       ['activation_6[0][0]']           \n",
      "                                                                                                  \n",
      " batch_normalization_10 (BatchN  (None, 32, 32, 128)  512        ['conv2d_9[0][0]']               \n",
      " ormalization)                                                                                    \n",
      "                                                                                                  \n",
      " conv2d_11 (Conv2D)             (None, 32, 32, 128)  8320        ['max_pooling2d_2[0][0]']        \n",
      "                                                                                                  \n",
      " activation_7 (Activation)      (None, 32, 32, 128)  0           ['batch_normalization_10[0][0]'] \n",
      "                                                                                                  \n",
      " batch_normalization_11 (BatchN  (None, 32, 32, 128)  512        ['conv2d_11[0][0]']              \n",
      " ormalization)                                                                                    \n",
      "                                                                                                  \n",
      " conv2d_10 (Conv2D)             (None, 32, 32, 128)  147584      ['activation_7[0][0]']           \n",
      "                                                                                                  \n",
      " add_3 (Add)                    (None, 32, 32, 128)  0           ['batch_normalization_11[0][0]', \n",
      "                                                                  'conv2d_10[0][0]']              \n",
      "                                                                                                  \n",
      " up_sampling2d (UpSampling2D)   (None, 64, 64, 128)  0           ['add_3[0][0]']                  \n",
      "                                                                                                  \n",
      " concatenate (Concatenate)      (None, 64, 64, 192)  0           ['up_sampling2d[0][0]',          \n",
      "                                                                  'add_2[0][0]']                  \n",
      "                                                                                                  \n",
      " conv2d_12 (Conv2D)             (None, 64, 64, 64)   4160        ['add_2[0][0]']                  \n",
      "                                                                                                  \n",
      " conv2d_13 (Conv2D)             (None, 64, 64, 64)   12352       ['concatenate[0][0]']            \n",
      "                                                                                                  \n",
      " batch_normalization_12 (BatchN  (None, 64, 64, 64)  256         ['conv2d_12[0][0]']              \n",
      " ormalization)                                                                                    \n",
      "                                                                                                  \n",
      " batch_normalization_13 (BatchN  (None, 64, 64, 64)  256         ['conv2d_13[0][0]']              \n",
      " ormalization)                                                                                    \n",
      "                                                                                                  \n",
      " add_4 (Add)                    (None, 64, 64, 64)   0           ['batch_normalization_12[0][0]', \n",
      "                                                                  'batch_normalization_13[0][0]'] \n",
      "                                                                                                  \n",
      " activation_8 (Activation)      (None, 64, 64, 64)   0           ['add_4[0][0]']                  \n",
      "                                                                                                  \n",
      " conv2d_14 (Conv2D)             (None, 64, 64, 1)    65          ['activation_8[0][0]']           \n",
      "                                                                                                  \n",
      " activation_9 (Activation)      (None, 64, 64, 1)    0           ['conv2d_14[0][0]']              \n",
      "                                                                                                  \n",
      " multiply (Multiply)            (None, 64, 64, 192)  0           ['concatenate[0][0]',            \n",
      "                                                                  'activation_9[0][0]']           \n",
      "                                                                                                  \n",
      " batch_normalization_14 (BatchN  (None, 64, 64, 192)  768        ['multiply[0][0]']               \n",
      " ormalization)                                                                                    \n",
      "                                                                                                  \n",
      " activation_10 (Activation)     (None, 64, 64, 192)  0           ['batch_normalization_14[0][0]'] \n",
      "                                                                                                  \n",
      " conv2d_15 (Conv2D)             (None, 64, 64, 64)   110656      ['activation_10[0][0]']          \n",
      "                                                                                                  \n",
      " batch_normalization_15 (BatchN  (None, 64, 64, 64)  256         ['conv2d_15[0][0]']              \n",
      " ormalization)                                                                                    \n",
      "                                                                                                  \n",
      " conv2d_17 (Conv2D)             (None, 64, 64, 64)   12352       ['multiply[0][0]']               \n",
      "                                                                                                  \n",
      " activation_11 (Activation)     (None, 64, 64, 64)   0           ['batch_normalization_15[0][0]'] \n",
      "                                                                                                  \n",
      " batch_normalization_16 (BatchN  (None, 64, 64, 64)  256         ['conv2d_17[0][0]']              \n",
      " ormalization)                                                                                    \n",
      "                                                                                                  \n",
      " conv2d_16 (Conv2D)             (None, 64, 64, 64)   36928       ['activation_11[0][0]']          \n",
      "                                                                                                  \n",
      " add_5 (Add)                    (None, 64, 64, 64)   0           ['batch_normalization_16[0][0]', \n",
      "                                                                  'conv2d_16[0][0]']              \n",
      "                                                                                                  \n",
      " up_sampling2d_1 (UpSampling2D)  (None, 128, 128, 64  0          ['add_5[0][0]']                  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " concatenate_1 (Concatenate)    (None, 128, 128, 96  0           ['up_sampling2d_1[0][0]',        \n",
      "                                )                                 'add_1[0][0]']                  \n",
      "                                                                                                  \n",
      " conv2d_18 (Conv2D)             (None, 128, 128, 32  1056        ['add_1[0][0]']                  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_19 (Conv2D)             (None, 128, 128, 32  3104        ['concatenate_1[0][0]']          \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_17 (BatchN  (None, 128, 128, 32  128        ['conv2d_18[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_18 (BatchN  (None, 128, 128, 32  128        ['conv2d_19[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " add_6 (Add)                    (None, 128, 128, 32  0           ['batch_normalization_17[0][0]', \n",
      "                                )                                 'batch_normalization_18[0][0]'] \n",
      "                                                                                                  \n",
      " activation_12 (Activation)     (None, 128, 128, 32  0           ['add_6[0][0]']                  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_20 (Conv2D)             (None, 128, 128, 1)  33          ['activation_12[0][0]']          \n",
      "                                                                                                  \n",
      " activation_13 (Activation)     (None, 128, 128, 1)  0           ['conv2d_20[0][0]']              \n",
      "                                                                                                  \n",
      " multiply_1 (Multiply)          (None, 128, 128, 96  0           ['concatenate_1[0][0]',          \n",
      "                                )                                 'activation_13[0][0]']          \n",
      "                                                                                                  \n",
      " batch_normalization_19 (BatchN  (None, 128, 128, 96  384        ['multiply_1[0][0]']             \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " activation_14 (Activation)     (None, 128, 128, 96  0           ['batch_normalization_19[0][0]'] \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_21 (Conv2D)             (None, 128, 128, 32  27680       ['activation_14[0][0]']          \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_20 (BatchN  (None, 128, 128, 32  128        ['conv2d_21[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_23 (Conv2D)             (None, 128, 128, 32  3104        ['multiply_1[0][0]']             \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " activation_15 (Activation)     (None, 128, 128, 32  0           ['batch_normalization_20[0][0]'] \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_21 (BatchN  (None, 128, 128, 32  128        ['conv2d_23[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_22 (Conv2D)             (None, 128, 128, 32  9248        ['activation_15[0][0]']          \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " add_7 (Add)                    (None, 128, 128, 32  0           ['batch_normalization_21[0][0]', \n",
      "                                )                                 'conv2d_22[0][0]']              \n",
      "                                                                                                  \n",
      " up_sampling2d_2 (UpSampling2D)  (None, 256, 256, 32  0          ['add_7[0][0]']                  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " concatenate_2 (Concatenate)    (None, 256, 256, 48  0           ['up_sampling2d_2[0][0]',        \n",
      "                                )                                 'add[0][0]']                    \n",
      "                                                                                                  \n",
      " conv2d_24 (Conv2D)             (None, 256, 256, 16  272         ['add[0][0]']                    \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_25 (Conv2D)             (None, 256, 256, 16  784         ['concatenate_2[0][0]']          \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_22 (BatchN  (None, 256, 256, 16  64         ['conv2d_24[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_23 (BatchN  (None, 256, 256, 16  64         ['conv2d_25[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " add_8 (Add)                    (None, 256, 256, 16  0           ['batch_normalization_22[0][0]', \n",
      "                                )                                 'batch_normalization_23[0][0]'] \n",
      "                                                                                                  \n",
      " activation_16 (Activation)     (None, 256, 256, 16  0           ['add_8[0][0]']                  \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_26 (Conv2D)             (None, 256, 256, 1)  17          ['activation_16[0][0]']          \n",
      "                                                                                                  \n",
      " activation_17 (Activation)     (None, 256, 256, 1)  0           ['conv2d_26[0][0]']              \n",
      "                                                                                                  \n",
      " multiply_2 (Multiply)          (None, 256, 256, 48  0           ['concatenate_2[0][0]',          \n",
      "                                )                                 'activation_17[0][0]']          \n",
      "                                                                                                  \n",
      " batch_normalization_24 (BatchN  (None, 256, 256, 48  192        ['multiply_2[0][0]']             \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " activation_18 (Activation)     (None, 256, 256, 48  0           ['batch_normalization_24[0][0]'] \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_27 (Conv2D)             (None, 256, 256, 16  6928        ['activation_18[0][0]']          \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_25 (BatchN  (None, 256, 256, 16  64         ['conv2d_27[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_29 (Conv2D)             (None, 256, 256, 16  784         ['multiply_2[0][0]']             \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " activation_19 (Activation)     (None, 256, 256, 16  0           ['batch_normalization_25[0][0]'] \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " batch_normalization_26 (BatchN  (None, 256, 256, 16  64         ['conv2d_29[0][0]']              \n",
      " ormalization)                  )                                                                 \n",
      "                                                                                                  \n",
      " conv2d_28 (Conv2D)             (None, 256, 256, 16  2320        ['activation_19[0][0]']          \n",
      "                                )                                                                 \n",
      "                                                                                                  \n",
      " add_9 (Add)                    (None, 256, 256, 16  0           ['batch_normalization_26[0][0]', \n",
      "                                )                                 'conv2d_28[0][0]']              \n",
      "                                                                                                  \n",
      " conv2d_30 (Conv2D)             (None, 256, 256, 1)  17          ['add_9[0][0]']                  \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 541,608\n",
      "Trainable params: 538,854\n",
      "Non-trainable params: 2,754\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = residual_attention_unet()\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1245 images belonging to 1 classes.\n",
      "Found 1245 images belonging to 1 classes.\n",
      "Epoch 1/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0585 - iou_coefficient: 0.8919 - dice_coefficient: 0.9416 - accuracy: 0.9255Found 357 images belonging to 1 classes.\n",
      "Found 357 images belonging to 1 classes.\n",
      "\n",
      "Epoch 1: loss improved from inf to 0.05846, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 39s 359ms/step - loss: 0.0585 - iou_coefficient: 0.8919 - dice_coefficient: 0.9416 - accuracy: 0.9255 - val_loss: 0.1317 - val_iou_coefficient: 0.7670 - val_dice_coefficient: 0.8681 - val_accuracy: 0.8220\n",
      "Epoch 2/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0299 - iou_coefficient: 0.9419 - dice_coefficient: 0.9701 - accuracy: 0.9536\n",
      "Epoch 2: loss improved from 0.05846 to 0.02994, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0299 - iou_coefficient: 0.9419 - dice_coefficient: 0.9701 - accuracy: 0.9536 - val_loss: 0.1033 - val_iou_coefficient: 0.8126 - val_dice_coefficient: 0.8966 - val_accuracy: 0.8220\n",
      "Epoch 3/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0267 - iou_coefficient: 0.9480 - dice_coefficient: 0.9733 - accuracy: 0.9579\n",
      "Epoch 3: loss improved from 0.02994 to 0.02670, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0267 - iou_coefficient: 0.9480 - dice_coefficient: 0.9733 - accuracy: 0.9579 - val_loss: 0.0987 - val_iou_coefficient: 0.8205 - val_dice_coefficient: 0.9014 - val_accuracy: 0.8224\n",
      "Epoch 4/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0251 - iou_coefficient: 0.9510 - dice_coefficient: 0.9749 - accuracy: 0.9600\n",
      "Epoch 4: loss improved from 0.02670 to 0.02511, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 257ms/step - loss: 0.0251 - iou_coefficient: 0.9510 - dice_coefficient: 0.9749 - accuracy: 0.9600 - val_loss: 0.0803 - val_iou_coefficient: 0.8512 - val_dice_coefficient: 0.9196 - val_accuracy: 0.8565\n",
      "Epoch 5/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0242 - iou_coefficient: 0.9528 - dice_coefficient: 0.9758 - accuracy: 0.9612\n",
      "Epoch 5: loss improved from 0.02511 to 0.02418, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 279ms/step - loss: 0.0242 - iou_coefficient: 0.9528 - dice_coefficient: 0.9758 - accuracy: 0.9612 - val_loss: 0.0556 - val_iou_coefficient: 0.8947 - val_dice_coefficient: 0.9444 - val_accuracy: 0.9047\n",
      "Epoch 6/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0238 - iou_coefficient: 0.9536 - dice_coefficient: 0.9762 - accuracy: 0.9618\n",
      "Epoch 6: loss improved from 0.02418 to 0.02376, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 283ms/step - loss: 0.0238 - iou_coefficient: 0.9536 - dice_coefficient: 0.9762 - accuracy: 0.9618 - val_loss: 0.0439 - val_iou_coefficient: 0.9151 - val_dice_coefficient: 0.9556 - val_accuracy: 0.9265\n",
      "Epoch 7/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0236 - iou_coefficient: 0.9538 - dice_coefficient: 0.9764 - accuracy: 0.9618\n",
      "Epoch 7: loss improved from 0.02376 to 0.02363, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 283ms/step - loss: 0.0236 - iou_coefficient: 0.9538 - dice_coefficient: 0.9764 - accuracy: 0.9618 - val_loss: 0.0281 - val_iou_coefficient: 0.9452 - val_dice_coefficient: 0.9718 - val_accuracy: 0.9543\n",
      "Epoch 8/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0229 - iou_coefficient: 0.9551 - dice_coefficient: 0.9771 - accuracy: 0.9629\n",
      "Epoch 8: loss improved from 0.02363 to 0.02295, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 286ms/step - loss: 0.0229 - iou_coefficient: 0.9551 - dice_coefficient: 0.9771 - accuracy: 0.9629 - val_loss: 0.0259 - val_iou_coefficient: 0.9498 - val_dice_coefficient: 0.9743 - val_accuracy: 0.9575\n",
      "Epoch 9/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0226 - iou_coefficient: 0.9558 - dice_coefficient: 0.9774 - accuracy: 0.9633\n",
      "Epoch 9: loss improved from 0.02295 to 0.02260, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 275ms/step - loss: 0.0226 - iou_coefficient: 0.9558 - dice_coefficient: 0.9774 - accuracy: 0.9633 - val_loss: 0.0218 - val_iou_coefficient: 0.9576 - val_dice_coefficient: 0.9783 - val_accuracy: 0.9648\n",
      "Epoch 10/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0224 - iou_coefficient: 0.9561 - dice_coefficient: 0.9775 - accuracy: 0.9636\n",
      "Epoch 10: loss improved from 0.02260 to 0.02245, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 261ms/step - loss: 0.0224 - iou_coefficient: 0.9561 - dice_coefficient: 0.9775 - accuracy: 0.9636 - val_loss: 0.0212 - val_iou_coefficient: 0.9586 - val_dice_coefficient: 0.9789 - val_accuracy: 0.9659\n",
      "Epoch 11/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0222 - iou_coefficient: 0.9566 - dice_coefficient: 0.9778 - accuracy: 0.9639\n",
      "Epoch 11: loss improved from 0.02245 to 0.02221, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 261ms/step - loss: 0.0222 - iou_coefficient: 0.9566 - dice_coefficient: 0.9778 - accuracy: 0.9639 - val_loss: 0.0213 - val_iou_coefficient: 0.9585 - val_dice_coefficient: 0.9788 - val_accuracy: 0.9657\n",
      "Epoch 12/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0222 - iou_coefficient: 0.9565 - dice_coefficient: 0.9778 - accuracy: 0.9638\n",
      "Epoch 12: loss did not improve from 0.02221\n",
      "78/78 [==============================] - 20s 261ms/step - loss: 0.0222 - iou_coefficient: 0.9565 - dice_coefficient: 0.9778 - accuracy: 0.9638 - val_loss: 0.0213 - val_iou_coefficient: 0.9577 - val_dice_coefficient: 0.9784 - val_accuracy: 0.9653\n",
      "Epoch 13/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0224 - iou_coefficient: 0.9561 - dice_coefficient: 0.9776 - accuracy: 0.9635\n",
      "Epoch 13: loss did not improve from 0.02221\n",
      "78/78 [==============================] - 21s 264ms/step - loss: 0.0224 - iou_coefficient: 0.9561 - dice_coefficient: 0.9776 - accuracy: 0.9635 - val_loss: 0.0208 - val_iou_coefficient: 0.9593 - val_dice_coefficient: 0.9792 - val_accuracy: 0.9661\n",
      "Epoch 14/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0218 - iou_coefficient: 0.9573 - dice_coefficient: 0.9782 - accuracy: 0.9644\n",
      "Epoch 14: loss improved from 0.02221 to 0.02185, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 275ms/step - loss: 0.0218 - iou_coefficient: 0.9573 - dice_coefficient: 0.9782 - accuracy: 0.9644 - val_loss: 0.0206 - val_iou_coefficient: 0.9595 - val_dice_coefficient: 0.9793 - val_accuracy: 0.9666\n",
      "Epoch 15/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0217 - iou_coefficient: 0.9576 - dice_coefficient: 0.9783 - accuracy: 0.9647\n",
      "Epoch 15: loss improved from 0.02185 to 0.02167, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 262ms/step - loss: 0.0217 - iou_coefficient: 0.9576 - dice_coefficient: 0.9783 - accuracy: 0.9647 - val_loss: 0.0205 - val_iou_coefficient: 0.9597 - val_dice_coefficient: 0.9794 - val_accuracy: 0.9665\n",
      "Epoch 16/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0218 - iou_coefficient: 0.9574 - dice_coefficient: 0.9782 - accuracy: 0.9645\n",
      "Epoch 16: loss did not improve from 0.02167\n",
      "78/78 [==============================] - 21s 275ms/step - loss: 0.0218 - iou_coefficient: 0.9574 - dice_coefficient: 0.9782 - accuracy: 0.9645 - val_loss: 0.0207 - val_iou_coefficient: 0.9596 - val_dice_coefficient: 0.9794 - val_accuracy: 0.9662\n",
      "Epoch 17/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0219 - iou_coefficient: 0.9572 - dice_coefficient: 0.9781 - accuracy: 0.9643\n",
      "Epoch 17: loss did not improve from 0.02167\n",
      "78/78 [==============================] - 20s 256ms/step - loss: 0.0219 - iou_coefficient: 0.9572 - dice_coefficient: 0.9781 - accuracy: 0.9643 - val_loss: 0.0217 - val_iou_coefficient: 0.9575 - val_dice_coefficient: 0.9783 - val_accuracy: 0.9651\n",
      "Epoch 18/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0213 - iou_coefficient: 0.9582 - dice_coefficient: 0.9787 - accuracy: 0.9651\n",
      "Epoch 18: loss improved from 0.02167 to 0.02134, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 264ms/step - loss: 0.0213 - iou_coefficient: 0.9582 - dice_coefficient: 0.9787 - accuracy: 0.9651 - val_loss: 0.0202 - val_iou_coefficient: 0.9604 - val_dice_coefficient: 0.9798 - val_accuracy: 0.9670\n",
      "Epoch 19/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0212 - iou_coefficient: 0.9584 - dice_coefficient: 0.9788 - accuracy: 0.9653\n",
      "Epoch 19: loss improved from 0.02134 to 0.02121, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 262ms/step - loss: 0.0212 - iou_coefficient: 0.9584 - dice_coefficient: 0.9788 - accuracy: 0.9653 - val_loss: 0.0196 - val_iou_coefficient: 0.9616 - val_dice_coefficient: 0.9804 - val_accuracy: 0.9681\n",
      "Epoch 20/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0213 - iou_coefficient: 0.9584 - dice_coefficient: 0.9787 - accuracy: 0.9653\n",
      "Epoch 20: loss did not improve from 0.02121\n",
      "78/78 [==============================] - 20s 254ms/step - loss: 0.0213 - iou_coefficient: 0.9584 - dice_coefficient: 0.9787 - accuracy: 0.9653 - val_loss: 0.0203 - val_iou_coefficient: 0.9600 - val_dice_coefficient: 0.9796 - val_accuracy: 0.9667\n",
      "Epoch 21/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0211 - iou_coefficient: 0.9586 - dice_coefficient: 0.9789 - accuracy: 0.9654\n",
      "Epoch 21: loss improved from 0.02121 to 0.02113, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 256ms/step - loss: 0.0211 - iou_coefficient: 0.9586 - dice_coefficient: 0.9789 - accuracy: 0.9654 - val_loss: 0.0199 - val_iou_coefficient: 0.9609 - val_dice_coefficient: 0.9800 - val_accuracy: 0.9676\n",
      "Epoch 22/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0212 - iou_coefficient: 0.9584 - dice_coefficient: 0.9788 - accuracy: 0.9652\n",
      "Epoch 22: loss did not improve from 0.02113\n",
      "78/78 [==============================] - 20s 253ms/step - loss: 0.0212 - iou_coefficient: 0.9584 - dice_coefficient: 0.9788 - accuracy: 0.9652 - val_loss: 0.0206 - val_iou_coefficient: 0.9591 - val_dice_coefficient: 0.9791 - val_accuracy: 0.9662\n",
      "Epoch 23/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0209 - iou_coefficient: 0.9590 - dice_coefficient: 0.9791 - accuracy: 0.9658\n",
      "Epoch 23: loss improved from 0.02113 to 0.02090, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 255ms/step - loss: 0.0209 - iou_coefficient: 0.9590 - dice_coefficient: 0.9791 - accuracy: 0.9658 - val_loss: 0.0218 - val_iou_coefficient: 0.9569 - val_dice_coefficient: 0.9780 - val_accuracy: 0.9639\n",
      "Epoch 24/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0209 - iou_coefficient: 0.9591 - dice_coefficient: 0.9791 - accuracy: 0.9658\n",
      "Epoch 24: loss improved from 0.02090 to 0.02087, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 256ms/step - loss: 0.0209 - iou_coefficient: 0.9591 - dice_coefficient: 0.9791 - accuracy: 0.9658 - val_loss: 0.0194 - val_iou_coefficient: 0.9618 - val_dice_coefficient: 0.9805 - val_accuracy: 0.9682\n",
      "Epoch 25/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0208 - iou_coefficient: 0.9593 - dice_coefficient: 0.9792 - accuracy: 0.9660\n",
      "Epoch 25: loss improved from 0.02087 to 0.02078, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 255ms/step - loss: 0.0208 - iou_coefficient: 0.9593 - dice_coefficient: 0.9792 - accuracy: 0.9660 - val_loss: 0.0206 - val_iou_coefficient: 0.9597 - val_dice_coefficient: 0.9794 - val_accuracy: 0.9663\n",
      "Epoch 26/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0206 - iou_coefficient: 0.9596 - dice_coefficient: 0.9794 - accuracy: 0.9662\n",
      "Epoch 26: loss improved from 0.02078 to 0.02064, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 257ms/step - loss: 0.0206 - iou_coefficient: 0.9596 - dice_coefficient: 0.9794 - accuracy: 0.9662 - val_loss: 0.0198 - val_iou_coefficient: 0.9612 - val_dice_coefficient: 0.9802 - val_accuracy: 0.9675\n",
      "Epoch 27/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0204 - iou_coefficient: 0.9599 - dice_coefficient: 0.9796 - accuracy: 0.9665\n",
      "Epoch 27: loss improved from 0.02064 to 0.02044, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 269ms/step - loss: 0.0204 - iou_coefficient: 0.9599 - dice_coefficient: 0.9796 - accuracy: 0.9665 - val_loss: 0.0210 - val_iou_coefficient: 0.9590 - val_dice_coefficient: 0.9791 - val_accuracy: 0.9658\n",
      "Epoch 28/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0208 - iou_coefficient: 0.9593 - dice_coefficient: 0.9792 - accuracy: 0.9660\n",
      "Epoch 28: loss did not improve from 0.02044\n",
      "78/78 [==============================] - 22s 279ms/step - loss: 0.0208 - iou_coefficient: 0.9593 - dice_coefficient: 0.9792 - accuracy: 0.9660 - val_loss: 0.0204 - val_iou_coefficient: 0.9600 - val_dice_coefficient: 0.9796 - val_accuracy: 0.9665\n",
      "Epoch 29/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0205 - iou_coefficient: 0.9599 - dice_coefficient: 0.9795 - accuracy: 0.9665\n",
      "Epoch 29: loss did not improve from 0.02044\n",
      "78/78 [==============================] - 21s 272ms/step - loss: 0.0205 - iou_coefficient: 0.9599 - dice_coefficient: 0.9795 - accuracy: 0.9665 - val_loss: 0.0199 - val_iou_coefficient: 0.9610 - val_dice_coefficient: 0.9801 - val_accuracy: 0.9672\n",
      "Epoch 30/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0204 - iou_coefficient: 0.9600 - dice_coefficient: 0.9796 - accuracy: 0.9665\n",
      "Epoch 30: loss improved from 0.02044 to 0.02043, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 275ms/step - loss: 0.0204 - iou_coefficient: 0.9600 - dice_coefficient: 0.9796 - accuracy: 0.9665 - val_loss: 0.0199 - val_iou_coefficient: 0.9608 - val_dice_coefficient: 0.9800 - val_accuracy: 0.9673\n",
      "Epoch 31/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0202 - iou_coefficient: 0.9604 - dice_coefficient: 0.9798 - accuracy: 0.9669\n",
      "Epoch 31: loss improved from 0.02043 to 0.02022, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 270ms/step - loss: 0.0202 - iou_coefficient: 0.9604 - dice_coefficient: 0.9798 - accuracy: 0.9669 - val_loss: 0.0199 - val_iou_coefficient: 0.9610 - val_dice_coefficient: 0.9801 - val_accuracy: 0.9672\n",
      "Epoch 32/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0201 - iou_coefficient: 0.9606 - dice_coefficient: 0.9799 - accuracy: 0.9670\n",
      "Epoch 32: loss improved from 0.02022 to 0.02012, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 278ms/step - loss: 0.0201 - iou_coefficient: 0.9606 - dice_coefficient: 0.9799 - accuracy: 0.9670 - val_loss: 0.0219 - val_iou_coefficient: 0.9572 - val_dice_coefficient: 0.9781 - val_accuracy: 0.9637\n",
      "Epoch 33/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0202 - iou_coefficient: 0.9604 - dice_coefficient: 0.9798 - accuracy: 0.9669\n",
      "Epoch 33: loss did not improve from 0.02012\n",
      "78/78 [==============================] - 21s 274ms/step - loss: 0.0202 - iou_coefficient: 0.9604 - dice_coefficient: 0.9798 - accuracy: 0.9669 - val_loss: 0.0198 - val_iou_coefficient: 0.9611 - val_dice_coefficient: 0.9802 - val_accuracy: 0.9677\n",
      "Epoch 34/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0200 - iou_coefficient: 0.9608 - dice_coefficient: 0.9800 - accuracy: 0.9672\n",
      "Epoch 34: loss improved from 0.02012 to 0.02001, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 275ms/step - loss: 0.0200 - iou_coefficient: 0.9608 - dice_coefficient: 0.9800 - accuracy: 0.9672 - val_loss: 0.0186 - val_iou_coefficient: 0.9632 - val_dice_coefficient: 0.9813 - val_accuracy: 0.9695\n",
      "Epoch 35/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0199 - iou_coefficient: 0.9610 - dice_coefficient: 0.9801 - accuracy: 0.9674\n",
      "Epoch 35: loss improved from 0.02001 to 0.01987, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 273ms/step - loss: 0.0199 - iou_coefficient: 0.9610 - dice_coefficient: 0.9801 - accuracy: 0.9674 - val_loss: 0.0193 - val_iou_coefficient: 0.9624 - val_dice_coefficient: 0.9808 - val_accuracy: 0.9683\n",
      "Epoch 36/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0199 - iou_coefficient: 0.9610 - dice_coefficient: 0.9801 - accuracy: 0.9674\n",
      "Epoch 36: loss improved from 0.01987 to 0.01987, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 277ms/step - loss: 0.0199 - iou_coefficient: 0.9610 - dice_coefficient: 0.9801 - accuracy: 0.9674 - val_loss: 0.0201 - val_iou_coefficient: 0.9608 - val_dice_coefficient: 0.9800 - val_accuracy: 0.9669\n",
      "Epoch 37/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0198 - iou_coefficient: 0.9611 - dice_coefficient: 0.9802 - accuracy: 0.9675\n",
      "Epoch 37: loss improved from 0.01987 to 0.01981, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 274ms/step - loss: 0.0198 - iou_coefficient: 0.9611 - dice_coefficient: 0.9802 - accuracy: 0.9675 - val_loss: 0.0193 - val_iou_coefficient: 0.9623 - val_dice_coefficient: 0.9808 - val_accuracy: 0.9683\n",
      "Epoch 38/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0198 - iou_coefficient: 0.9611 - dice_coefficient: 0.9802 - accuracy: 0.9675\n",
      "Epoch 38: loss improved from 0.01981 to 0.01981, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 279ms/step - loss: 0.0198 - iou_coefficient: 0.9611 - dice_coefficient: 0.9802 - accuracy: 0.9675 - val_loss: 0.0208 - val_iou_coefficient: 0.9595 - val_dice_coefficient: 0.9793 - val_accuracy: 0.9655\n",
      "Epoch 39/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0196 - iou_coefficient: 0.9616 - dice_coefficient: 0.9804 - accuracy: 0.9678\n",
      "Epoch 39: loss improved from 0.01981 to 0.01960, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 278ms/step - loss: 0.0196 - iou_coefficient: 0.9616 - dice_coefficient: 0.9804 - accuracy: 0.9678 - val_loss: 0.0203 - val_iou_coefficient: 0.9605 - val_dice_coefficient: 0.9799 - val_accuracy: 0.9664\n",
      "Epoch 40/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0195 - iou_coefficient: 0.9617 - dice_coefficient: 0.9805 - accuracy: 0.9679\n",
      "Epoch 40: loss improved from 0.01960 to 0.01954, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 280ms/step - loss: 0.0195 - iou_coefficient: 0.9617 - dice_coefficient: 0.9805 - accuracy: 0.9679 - val_loss: 0.0194 - val_iou_coefficient: 0.9616 - val_dice_coefficient: 0.9804 - val_accuracy: 0.9679\n",
      "Epoch 41/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0196 - iou_coefficient: 0.9615 - dice_coefficient: 0.9804 - accuracy: 0.9678\n",
      "Epoch 41: loss did not improve from 0.01954\n",
      "78/78 [==============================] - 21s 265ms/step - loss: 0.0196 - iou_coefficient: 0.9615 - dice_coefficient: 0.9804 - accuracy: 0.9678 - val_loss: 0.0199 - val_iou_coefficient: 0.9604 - val_dice_coefficient: 0.9798 - val_accuracy: 0.9670\n",
      "Epoch 42/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0196 - iou_coefficient: 0.9616 - dice_coefficient: 0.9804 - accuracy: 0.9679\n",
      "Epoch 42: loss did not improve from 0.01954\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0196 - iou_coefficient: 0.9616 - dice_coefficient: 0.9804 - accuracy: 0.9679 - val_loss: 0.0195 - val_iou_coefficient: 0.9619 - val_dice_coefficient: 0.9806 - val_accuracy: 0.9680\n",
      "Epoch 43/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0196 - iou_coefficient: 0.9616 - dice_coefficient: 0.9804 - accuracy: 0.9679\n",
      "Epoch 43: loss did not improve from 0.01954\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0196 - iou_coefficient: 0.9616 - dice_coefficient: 0.9804 - accuracy: 0.9679 - val_loss: 0.0194 - val_iou_coefficient: 0.9614 - val_dice_coefficient: 0.9803 - val_accuracy: 0.9680\n",
      "Epoch 44/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0195 - iou_coefficient: 0.9618 - dice_coefficient: 0.9805 - accuracy: 0.9680\n",
      "Epoch 44: loss improved from 0.01954 to 0.01950, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 277ms/step - loss: 0.0195 - iou_coefficient: 0.9618 - dice_coefficient: 0.9805 - accuracy: 0.9680 - val_loss: 0.0191 - val_iou_coefficient: 0.9624 - val_dice_coefficient: 0.9808 - val_accuracy: 0.9685\n",
      "Epoch 45/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0191 - iou_coefficient: 0.9625 - dice_coefficient: 0.9809 - accuracy: 0.9686\n",
      "Epoch 45: loss improved from 0.01950 to 0.01912, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 285ms/step - loss: 0.0191 - iou_coefficient: 0.9625 - dice_coefficient: 0.9809 - accuracy: 0.9686 - val_loss: 0.0207 - val_iou_coefficient: 0.9595 - val_dice_coefficient: 0.9793 - val_accuracy: 0.9658\n",
      "Epoch 46/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0193 - iou_coefficient: 0.9621 - dice_coefficient: 0.9807 - accuracy: 0.9683\n",
      "Epoch 46: loss did not improve from 0.01912\n",
      "78/78 [==============================] - 20s 259ms/step - loss: 0.0193 - iou_coefficient: 0.9621 - dice_coefficient: 0.9807 - accuracy: 0.9683 - val_loss: 0.0192 - val_iou_coefficient: 0.9625 - val_dice_coefficient: 0.9809 - val_accuracy: 0.9684\n",
      "Epoch 47/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0195 - iou_coefficient: 0.9618 - dice_coefficient: 0.9805 - accuracy: 0.9681\n",
      "Epoch 47: loss did not improve from 0.01912\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0195 - iou_coefficient: 0.9618 - dice_coefficient: 0.9805 - accuracy: 0.9681 - val_loss: 0.0205 - val_iou_coefficient: 0.9603 - val_dice_coefficient: 0.9797 - val_accuracy: 0.9661\n",
      "Epoch 48/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0193 - iou_coefficient: 0.9622 - dice_coefficient: 0.9807 - accuracy: 0.9683\n",
      "Epoch 48: loss did not improve from 0.01912\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0193 - iou_coefficient: 0.9622 - dice_coefficient: 0.9807 - accuracy: 0.9683 - val_loss: 0.0194 - val_iou_coefficient: 0.9617 - val_dice_coefficient: 0.9805 - val_accuracy: 0.9681\n",
      "Epoch 49/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0190 - iou_coefficient: 0.9627 - dice_coefficient: 0.9810 - accuracy: 0.9688\n",
      "Epoch 49: loss improved from 0.01912 to 0.01900, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 262ms/step - loss: 0.0190 - iou_coefficient: 0.9627 - dice_coefficient: 0.9810 - accuracy: 0.9688 - val_loss: 0.0189 - val_iou_coefficient: 0.9630 - val_dice_coefficient: 0.9811 - val_accuracy: 0.9690\n",
      "Epoch 50/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0194 - iou_coefficient: 0.9621 - dice_coefficient: 0.9807 - accuracy: 0.9682\n",
      "Epoch 50: loss did not improve from 0.01900\n",
      "78/78 [==============================] - 20s 254ms/step - loss: 0.0194 - iou_coefficient: 0.9621 - dice_coefficient: 0.9807 - accuracy: 0.9682 - val_loss: 0.0204 - val_iou_coefficient: 0.9603 - val_dice_coefficient: 0.9797 - val_accuracy: 0.9662\n",
      "Epoch 51/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0191 - iou_coefficient: 0.9626 - dice_coefficient: 0.9809 - accuracy: 0.9687\n",
      "Epoch 51: loss did not improve from 0.01900\n",
      "78/78 [==============================] - 20s 259ms/step - loss: 0.0191 - iou_coefficient: 0.9626 - dice_coefficient: 0.9809 - accuracy: 0.9687 - val_loss: 0.0203 - val_iou_coefficient: 0.9604 - val_dice_coefficient: 0.9798 - val_accuracy: 0.9665\n",
      "Epoch 52/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0192 - iou_coefficient: 0.9623 - dice_coefficient: 0.9808 - accuracy: 0.9684\n",
      "Epoch 52: loss did not improve from 0.01900\n",
      "78/78 [==============================] - 20s 256ms/step - loss: 0.0192 - iou_coefficient: 0.9623 - dice_coefficient: 0.9808 - accuracy: 0.9684 - val_loss: 0.0192 - val_iou_coefficient: 0.9627 - val_dice_coefficient: 0.9810 - val_accuracy: 0.9683\n",
      "Epoch 53/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0190 - iou_coefficient: 0.9627 - dice_coefficient: 0.9810 - accuracy: 0.9688\n",
      "Epoch 53: loss did not improve from 0.01900\n",
      "78/78 [==============================] - 20s 259ms/step - loss: 0.0190 - iou_coefficient: 0.9627 - dice_coefficient: 0.9810 - accuracy: 0.9688 - val_loss: 0.0191 - val_iou_coefficient: 0.9623 - val_dice_coefficient: 0.9808 - val_accuracy: 0.9685\n",
      "Epoch 54/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0191 - iou_coefficient: 0.9625 - dice_coefficient: 0.9809 - accuracy: 0.9686\n",
      "Epoch 54: loss did not improve from 0.01900\n",
      "78/78 [==============================] - 20s 259ms/step - loss: 0.0191 - iou_coefficient: 0.9625 - dice_coefficient: 0.9809 - accuracy: 0.9686 - val_loss: 0.0195 - val_iou_coefficient: 0.9617 - val_dice_coefficient: 0.9805 - val_accuracy: 0.9678\n",
      "Epoch 55/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0188 - iou_coefficient: 0.9630 - dice_coefficient: 0.9812 - accuracy: 0.9691\n",
      "Epoch 55: loss improved from 0.01900 to 0.01883, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 266ms/step - loss: 0.0188 - iou_coefficient: 0.9630 - dice_coefficient: 0.9812 - accuracy: 0.9691 - val_loss: 0.0196 - val_iou_coefficient: 0.9616 - val_dice_coefficient: 0.9804 - val_accuracy: 0.9676\n",
      "Epoch 56/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0190 - iou_coefficient: 0.9628 - dice_coefficient: 0.9810 - accuracy: 0.9688\n",
      "Epoch 56: loss did not improve from 0.01883\n",
      "78/78 [==============================] - 20s 255ms/step - loss: 0.0190 - iou_coefficient: 0.9628 - dice_coefficient: 0.9810 - accuracy: 0.9688 - val_loss: 0.0202 - val_iou_coefficient: 0.9605 - val_dice_coefficient: 0.9799 - val_accuracy: 0.9667\n",
      "Epoch 57/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0190 - iou_coefficient: 0.9626 - dice_coefficient: 0.9810 - accuracy: 0.9687\n",
      "Epoch 57: loss did not improve from 0.01883\n",
      "78/78 [==============================] - 20s 261ms/step - loss: 0.0190 - iou_coefficient: 0.9626 - dice_coefficient: 0.9810 - accuracy: 0.9687 - val_loss: 0.0189 - val_iou_coefficient: 0.9632 - val_dice_coefficient: 0.9812 - val_accuracy: 0.9688\n",
      "Epoch 58/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0191 - iou_coefficient: 0.9625 - dice_coefficient: 0.9809 - accuracy: 0.9686\n",
      "Epoch 58: loss did not improve from 0.01883\n",
      "78/78 [==============================] - 20s 254ms/step - loss: 0.0191 - iou_coefficient: 0.9625 - dice_coefficient: 0.9809 - accuracy: 0.9686 - val_loss: 0.0204 - val_iou_coefficient: 0.9598 - val_dice_coefficient: 0.9795 - val_accuracy: 0.9667\n",
      "Epoch 59/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0189 - iou_coefficient: 0.9630 - dice_coefficient: 0.9811 - accuracy: 0.9690\n",
      "Epoch 59: loss did not improve from 0.01883\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0189 - iou_coefficient: 0.9630 - dice_coefficient: 0.9811 - accuracy: 0.9690 - val_loss: 0.0191 - val_iou_coefficient: 0.9628 - val_dice_coefficient: 0.9810 - val_accuracy: 0.9687\n",
      "Epoch 60/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0188 - iou_coefficient: 0.9632 - dice_coefficient: 0.9812 - accuracy: 0.9692\n",
      "Epoch 60: loss improved from 0.01883 to 0.01875, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 20s 262ms/step - loss: 0.0188 - iou_coefficient: 0.9632 - dice_coefficient: 0.9812 - accuracy: 0.9692 - val_loss: 0.0183 - val_iou_coefficient: 0.9639 - val_dice_coefficient: 0.9816 - val_accuracy: 0.9699\n",
      "Epoch 61/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0188 - iou_coefficient: 0.9630 - dice_coefficient: 0.9812 - accuracy: 0.9691\n",
      "Epoch 61: loss did not improve from 0.01875\n",
      "78/78 [==============================] - 20s 257ms/step - loss: 0.0188 - iou_coefficient: 0.9630 - dice_coefficient: 0.9812 - accuracy: 0.9691 - val_loss: 0.0203 - val_iou_coefficient: 0.9602 - val_dice_coefficient: 0.9797 - val_accuracy: 0.9664\n",
      "Epoch 62/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0186 - iou_coefficient: 0.9635 - dice_coefficient: 0.9814 - accuracy: 0.9695\n",
      "Epoch 62: loss improved from 0.01875 to 0.01859, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 274ms/step - loss: 0.0186 - iou_coefficient: 0.9635 - dice_coefficient: 0.9814 - accuracy: 0.9695 - val_loss: 0.0187 - val_iou_coefficient: 0.9632 - val_dice_coefficient: 0.9812 - val_accuracy: 0.9692\n",
      "Epoch 63/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0186 - iou_coefficient: 0.9634 - dice_coefficient: 0.9814 - accuracy: 0.9694\n",
      "Epoch 63: loss did not improve from 0.01859\n",
      "78/78 [==============================] - 21s 265ms/step - loss: 0.0186 - iou_coefficient: 0.9634 - dice_coefficient: 0.9814 - accuracy: 0.9694 - val_loss: 0.0195 - val_iou_coefficient: 0.9620 - val_dice_coefficient: 0.9806 - val_accuracy: 0.9676\n",
      "Epoch 64/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9814 - accuracy: 0.9695\n",
      "Epoch 64: loss improved from 0.01859 to 0.01855, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 266ms/step - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9814 - accuracy: 0.9695 - val_loss: 0.0189 - val_iou_coefficient: 0.9627 - val_dice_coefficient: 0.9810 - val_accuracy: 0.9687\n",
      "Epoch 65/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9815 - accuracy: 0.9696\n",
      "Epoch 65: loss improved from 0.01855 to 0.01853, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 274ms/step - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9815 - accuracy: 0.9696 - val_loss: 0.0199 - val_iou_coefficient: 0.9607 - val_dice_coefficient: 0.9799 - val_accuracy: 0.9670\n",
      "Epoch 66/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0186 - iou_coefficient: 0.9634 - dice_coefficient: 0.9814 - accuracy: 0.9694\n",
      "Epoch 66: loss did not improve from 0.01853\n",
      "78/78 [==============================] - 21s 273ms/step - loss: 0.0186 - iou_coefficient: 0.9634 - dice_coefficient: 0.9814 - accuracy: 0.9694 - val_loss: 0.0197 - val_iou_coefficient: 0.9618 - val_dice_coefficient: 0.9805 - val_accuracy: 0.9674\n",
      "Epoch 67/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9815 - accuracy: 0.9695\n",
      "Epoch 67: loss did not improve from 0.01853\n",
      "78/78 [==============================] - 20s 262ms/step - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9815 - accuracy: 0.9695 - val_loss: 0.0186 - val_iou_coefficient: 0.9633 - val_dice_coefficient: 0.9813 - val_accuracy: 0.9694\n",
      "Epoch 68/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9815 - accuracy: 0.9695\n",
      "Epoch 68: loss did not improve from 0.01853\n",
      "78/78 [==============================] - 21s 264ms/step - loss: 0.0185 - iou_coefficient: 0.9636 - dice_coefficient: 0.9815 - accuracy: 0.9695 - val_loss: 0.0187 - val_iou_coefficient: 0.9631 - val_dice_coefficient: 0.9812 - val_accuracy: 0.9692\n",
      "Epoch 69/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0185 - iou_coefficient: 0.9637 - dice_coefficient: 0.9815 - accuracy: 0.9696\n",
      "Epoch 69: loss improved from 0.01853 to 0.01849, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 271ms/step - loss: 0.0185 - iou_coefficient: 0.9637 - dice_coefficient: 0.9815 - accuracy: 0.9696 - val_loss: 0.0200 - val_iou_coefficient: 0.9606 - val_dice_coefficient: 0.9799 - val_accuracy: 0.9668\n",
      "Epoch 70/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0186 - iou_coefficient: 0.9635 - dice_coefficient: 0.9814 - accuracy: 0.9694\n",
      "Epoch 70: loss did not improve from 0.01849\n",
      "78/78 [==============================] - 20s 260ms/step - loss: 0.0186 - iou_coefficient: 0.9635 - dice_coefficient: 0.9814 - accuracy: 0.9694 - val_loss: 0.0192 - val_iou_coefficient: 0.9616 - val_dice_coefficient: 0.9804 - val_accuracy: 0.9684\n",
      "Epoch 71/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0184 - iou_coefficient: 0.9638 - dice_coefficient: 0.9816 - accuracy: 0.9698\n",
      "Epoch 71: loss improved from 0.01849 to 0.01841, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 270ms/step - loss: 0.0184 - iou_coefficient: 0.9638 - dice_coefficient: 0.9816 - accuracy: 0.9698 - val_loss: 0.0188 - val_iou_coefficient: 0.9628 - val_dice_coefficient: 0.9810 - val_accuracy: 0.9690\n",
      "Epoch 72/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0185 - iou_coefficient: 0.9637 - dice_coefficient: 0.9815 - accuracy: 0.9697\n",
      "Epoch 72: loss did not improve from 0.01841\n",
      "78/78 [==============================] - 21s 266ms/step - loss: 0.0185 - iou_coefficient: 0.9637 - dice_coefficient: 0.9815 - accuracy: 0.9697 - val_loss: 0.0182 - val_iou_coefficient: 0.9641 - val_dice_coefficient: 0.9817 - val_accuracy: 0.9700\n",
      "Epoch 73/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0184 - iou_coefficient: 0.9639 - dice_coefficient: 0.9816 - accuracy: 0.9698\n",
      "Epoch 73: loss improved from 0.01841 to 0.01838, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 269ms/step - loss: 0.0184 - iou_coefficient: 0.9639 - dice_coefficient: 0.9816 - accuracy: 0.9698 - val_loss: 0.0193 - val_iou_coefficient: 0.9618 - val_dice_coefficient: 0.9805 - val_accuracy: 0.9680\n",
      "Epoch 74/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0183 - iou_coefficient: 0.9641 - dice_coefficient: 0.9817 - accuracy: 0.9700\n",
      "Epoch 74: loss improved from 0.01838 to 0.01828, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 275ms/step - loss: 0.0183 - iou_coefficient: 0.9641 - dice_coefficient: 0.9817 - accuracy: 0.9700 - val_loss: 0.0189 - val_iou_coefficient: 0.9627 - val_dice_coefficient: 0.9810 - val_accuracy: 0.9689\n",
      "Epoch 75/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0183 - iou_coefficient: 0.9640 - dice_coefficient: 0.9817 - accuracy: 0.9699\n",
      "Epoch 75: loss did not improve from 0.01828\n",
      "78/78 [==============================] - 21s 267ms/step - loss: 0.0183 - iou_coefficient: 0.9640 - dice_coefficient: 0.9817 - accuracy: 0.9699 - val_loss: 0.0190 - val_iou_coefficient: 0.9629 - val_dice_coefficient: 0.9811 - val_accuracy: 0.9686\n",
      "Epoch 76/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0184 - iou_coefficient: 0.9639 - dice_coefficient: 0.9816 - accuracy: 0.9698\n",
      "Epoch 76: loss did not improve from 0.01828\n",
      "78/78 [==============================] - 21s 271ms/step - loss: 0.0184 - iou_coefficient: 0.9639 - dice_coefficient: 0.9816 - accuracy: 0.9698 - val_loss: 0.0184 - val_iou_coefficient: 0.9636 - val_dice_coefficient: 0.9815 - val_accuracy: 0.9697\n",
      "Epoch 77/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0181 - iou_coefficient: 0.9644 - dice_coefficient: 0.9819 - accuracy: 0.9702\n",
      "Epoch 77: loss improved from 0.01828 to 0.01812, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 268ms/step - loss: 0.0181 - iou_coefficient: 0.9644 - dice_coefficient: 0.9819 - accuracy: 0.9702 - val_loss: 0.0184 - val_iou_coefficient: 0.9638 - val_dice_coefficient: 0.9816 - val_accuracy: 0.9696\n",
      "Epoch 78/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0184 - iou_coefficient: 0.9638 - dice_coefficient: 0.9816 - accuracy: 0.9697\n",
      "Epoch 78: loss did not improve from 0.01812\n",
      "78/78 [==============================] - 21s 265ms/step - loss: 0.0184 - iou_coefficient: 0.9638 - dice_coefficient: 0.9816 - accuracy: 0.9697 - val_loss: 0.0193 - val_iou_coefficient: 0.9623 - val_dice_coefficient: 0.9808 - val_accuracy: 0.9681\n",
      "Epoch 79/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0183 - iou_coefficient: 0.9640 - dice_coefficient: 0.9817 - accuracy: 0.9699\n",
      "Epoch 79: loss did not improve from 0.01812\n",
      "78/78 [==============================] - 21s 270ms/step - loss: 0.0183 - iou_coefficient: 0.9640 - dice_coefficient: 0.9817 - accuracy: 0.9699 - val_loss: 0.0188 - val_iou_coefficient: 0.9624 - val_dice_coefficient: 0.9808 - val_accuracy: 0.9689\n",
      "Epoch 80/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0182 - iou_coefficient: 0.9643 - dice_coefficient: 0.9818 - accuracy: 0.9701\n",
      "Epoch 80: loss did not improve from 0.01812\n",
      "78/78 [==============================] - 20s 258ms/step - loss: 0.0182 - iou_coefficient: 0.9643 - dice_coefficient: 0.9818 - accuracy: 0.9701 - val_loss: 0.0196 - val_iou_coefficient: 0.9619 - val_dice_coefficient: 0.9806 - val_accuracy: 0.9676\n",
      "Epoch 81/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703\n",
      "Epoch 81: loss improved from 0.01812 to 0.01809, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 272ms/step - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703 - val_loss: 0.0178 - val_iou_coefficient: 0.9651 - val_dice_coefficient: 0.9822 - val_accuracy: 0.9706\n",
      "Epoch 82/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703\n",
      "Epoch 82: loss improved from 0.01809 to 0.01807, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 271ms/step - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703 - val_loss: 0.0194 - val_iou_coefficient: 0.9621 - val_dice_coefficient: 0.9807 - val_accuracy: 0.9681\n",
      "Epoch 83/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703\n",
      "Epoch 83: loss did not improve from 0.01807\n",
      "78/78 [==============================] - 20s 256ms/step - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703 - val_loss: 0.0188 - val_iou_coefficient: 0.9627 - val_dice_coefficient: 0.9810 - val_accuracy: 0.9688\n",
      "Epoch 84/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703\n",
      "Epoch 84: loss did not improve from 0.01807\n",
      "78/78 [==============================] - 20s 257ms/step - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703 - val_loss: 0.0196 - val_iou_coefficient: 0.9612 - val_dice_coefficient: 0.9802 - val_accuracy: 0.9675\n",
      "Epoch 85/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0180 - iou_coefficient: 0.9646 - dice_coefficient: 0.9820 - accuracy: 0.9704\n",
      "Epoch 85: loss improved from 0.01807 to 0.01803, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 263ms/step - loss: 0.0180 - iou_coefficient: 0.9646 - dice_coefficient: 0.9820 - accuracy: 0.9704 - val_loss: 0.0186 - val_iou_coefficient: 0.9635 - val_dice_coefficient: 0.9814 - val_accuracy: 0.9693\n",
      "Epoch 86/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703\n",
      "Epoch 86: loss did not improve from 0.01803\n",
      "78/78 [==============================] - 20s 262ms/step - loss: 0.0181 - iou_coefficient: 0.9645 - dice_coefficient: 0.9819 - accuracy: 0.9703 - val_loss: 0.0185 - val_iou_coefficient: 0.9640 - val_dice_coefficient: 0.9816 - val_accuracy: 0.9696\n",
      "Epoch 87/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0180 - iou_coefficient: 0.9646 - dice_coefficient: 0.9820 - accuracy: 0.9704\n",
      "Epoch 87: loss improved from 0.01803 to 0.01800, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 265ms/step - loss: 0.0180 - iou_coefficient: 0.9646 - dice_coefficient: 0.9820 - accuracy: 0.9704 - val_loss: 0.0191 - val_iou_coefficient: 0.9621 - val_dice_coefficient: 0.9807 - val_accuracy: 0.9684\n",
      "Epoch 88/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0180 - iou_coefficient: 0.9647 - dice_coefficient: 0.9820 - accuracy: 0.9705\n",
      "Epoch 88: loss improved from 0.01800 to 0.01797, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 267ms/step - loss: 0.0180 - iou_coefficient: 0.9647 - dice_coefficient: 0.9820 - accuracy: 0.9705 - val_loss: 0.0184 - val_iou_coefficient: 0.9640 - val_dice_coefficient: 0.9817 - val_accuracy: 0.9698\n",
      "Epoch 89/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709\n",
      "Epoch 89: loss improved from 0.01797 to 0.01773, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 269ms/step - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709 - val_loss: 0.0190 - val_iou_coefficient: 0.9624 - val_dice_coefficient: 0.9809 - val_accuracy: 0.9688\n",
      "Epoch 90/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0180 - iou_coefficient: 0.9646 - dice_coefficient: 0.9820 - accuracy: 0.9704\n",
      "Epoch 90: loss did not improve from 0.01773\n",
      "78/78 [==============================] - 21s 276ms/step - loss: 0.0180 - iou_coefficient: 0.9646 - dice_coefficient: 0.9820 - accuracy: 0.9704 - val_loss: 0.0193 - val_iou_coefficient: 0.9622 - val_dice_coefficient: 0.9807 - val_accuracy: 0.9680\n",
      "Epoch 91/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0180 - iou_coefficient: 0.9647 - dice_coefficient: 0.9820 - accuracy: 0.9705\n",
      "Epoch 91: loss did not improve from 0.01773\n",
      "78/78 [==============================] - 22s 279ms/step - loss: 0.0180 - iou_coefficient: 0.9647 - dice_coefficient: 0.9820 - accuracy: 0.9705 - val_loss: 0.0184 - val_iou_coefficient: 0.9642 - val_dice_coefficient: 0.9818 - val_accuracy: 0.9697\n",
      "Epoch 92/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0178 - iou_coefficient: 0.9651 - dice_coefficient: 0.9822 - accuracy: 0.9708\n",
      "Epoch 92: loss did not improve from 0.01773\n",
      "78/78 [==============================] - 22s 277ms/step - loss: 0.0178 - iou_coefficient: 0.9651 - dice_coefficient: 0.9822 - accuracy: 0.9708 - val_loss: 0.0190 - val_iou_coefficient: 0.9630 - val_dice_coefficient: 0.9812 - val_accuracy: 0.9685\n",
      "Epoch 93/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709\n",
      "Epoch 93: loss improved from 0.01773 to 0.01769, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 284ms/step - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709 - val_loss: 0.0193 - val_iou_coefficient: 0.9621 - val_dice_coefficient: 0.9807 - val_accuracy: 0.9682\n",
      "Epoch 94/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709\n",
      "Epoch 94: loss did not improve from 0.01769\n",
      "78/78 [==============================] - 21s 273ms/step - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709 - val_loss: 0.0184 - val_iou_coefficient: 0.9639 - val_dice_coefficient: 0.9816 - val_accuracy: 0.9697\n",
      "Epoch 95/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709\n",
      "Epoch 95: loss did not improve from 0.01769\n",
      "78/78 [==============================] - 21s 270ms/step - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709 - val_loss: 0.0186 - val_iou_coefficient: 0.9639 - val_dice_coefficient: 0.9816 - val_accuracy: 0.9694\n",
      "Epoch 96/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709\n",
      "Epoch 96: loss did not improve from 0.01769\n",
      "78/78 [==============================] - 22s 277ms/step - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709 - val_loss: 0.0184 - val_iou_coefficient: 0.9638 - val_dice_coefficient: 0.9816 - val_accuracy: 0.9698\n",
      "Epoch 97/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0177 - iou_coefficient: 0.9653 - dice_coefficient: 0.9823 - accuracy: 0.9709\n",
      "Epoch 97: loss improved from 0.01769 to 0.01768, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 22s 276ms/step - loss: 0.0177 - iou_coefficient: 0.9653 - dice_coefficient: 0.9823 - accuracy: 0.9709 - val_loss: 0.0179 - val_iou_coefficient: 0.9647 - val_dice_coefficient: 0.9820 - val_accuracy: 0.9705\n",
      "Epoch 98/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0178 - iou_coefficient: 0.9650 - dice_coefficient: 0.9822 - accuracy: 0.9708\n",
      "Epoch 98: loss did not improve from 0.01768\n",
      "78/78 [==============================] - 21s 269ms/step - loss: 0.0178 - iou_coefficient: 0.9650 - dice_coefficient: 0.9822 - accuracy: 0.9708 - val_loss: 0.0195 - val_iou_coefficient: 0.9614 - val_dice_coefficient: 0.9803 - val_accuracy: 0.9677\n",
      "Epoch 99/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709\n",
      "Epoch 99: loss did not improve from 0.01768\n",
      "78/78 [==============================] - 21s 275ms/step - loss: 0.0177 - iou_coefficient: 0.9652 - dice_coefficient: 0.9823 - accuracy: 0.9709 - val_loss: 0.0182 - val_iou_coefficient: 0.9646 - val_dice_coefficient: 0.9820 - val_accuracy: 0.9701\n",
      "Epoch 100/100\n",
      "78/78 [==============================] - ETA: 0s - loss: 0.0176 - iou_coefficient: 0.9654 - dice_coefficient: 0.9824 - accuracy: 0.9711\n",
      "Epoch 100: loss improved from 0.01768 to 0.01759, saving model to best_model.hdf5\n",
      "78/78 [==============================] - 21s 266ms/step - loss: 0.0176 - iou_coefficient: 0.9654 - dice_coefficient: 0.9824 - accuracy: 0.9711 - val_loss: 0.0187 - val_iou_coefficient: 0.9632 - val_dice_coefficient: 0.9813 - val_accuracy: 0.9693\n"
     ]
    }
   ],
   "source": [
    "model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=3e-4), loss=dice_loss, metrics=[iou_coefficient, dice_coefficient, \"accuracy\"])\n",
    "\n",
    "model_checkpoint = ModelCheckpoint(\"best_model.hdf5\", monitor=\"val_loss\", verbose=1, save_best_only=True)\n",
    "\n",
    "#每個時期迭代次數 = 資料夾圖片總數 / 訓練批次(batch_size)\n",
    "history = model.fit(myGene, epochs=100, steps_per_epoch=78, validation_data=valGene, validation_steps=23,\n",
    "                    callbacks=[model_checkpoint])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#保存訓練指標\n",
    "with open(\"trainHistory.txt\", \"wb\") as file_pi:\n",
    "    pickle.dump(history.history, file_pi)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#訓練指標視覺化\n",
    "fig, axs = plt.subplots(2, 2, figsize=(10, 10))\n",
    "\n",
    "axs[0, 0].plot(history.history[\"accuracy\"], label=\"Train\")\n",
    "axs[0, 0].plot(history.history[\"val_accuracy\"], label=\"Validation\")\n",
    "axs[0, 0].set_title(\"Accuracy\")\n",
    "axs[0, 0].set_xlabel(\"Epochs\")\n",
    "axs[0, 0].legend()\n",
    "\n",
    "axs[0, 1].plot(history.history[\"loss\"], label=\"Train\")\n",
    "axs[0, 1].plot(history.history[\"val_loss\"], label=\"Validation\")\n",
    "axs[0, 1].set_title(\"Dice Loss\")\n",
    "axs[0, 1].set_xlabel(\"Epochs\")\n",
    "axs[0, 1].legend()\n",
    "\n",
    "axs[1, 0].plot(history.history[\"dice_coefficient\"], label=\"Train\")\n",
    "axs[1, 0].plot(history.history[\"val_dice_coefficient\"], label=\"Validation\")\n",
    "axs[1, 0].set_title(\"Dice Coefficient\")\n",
    "axs[1, 0].set_xlabel(\"Epochs\")\n",
    "axs[1, 0].legend()\n",
    "\n",
    "axs[1, 1].plot(history.history[\"iou_coefficient\"], label=\"Train\")\n",
    "axs[1, 1].plot(history.history[\"val_iou_coefficient\"], label=\"Validation\")\n",
    "axs[1, 1].set_title(\"IOU Coefficient\")\n",
    "axs[1, 1].set_xlabel(\"Epochs\")\n",
    "axs[1, 1].legend()\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "180/180 [==============================] - 2s 11ms/step\n"
     ]
    }
   ],
   "source": [
    "#用訓練好的模型預測測試集圖像\n",
    "testGene = test_generator(r\"D:\\unet_ours\\data\\membrane\\test\")\n",
    "model.load_weights(\"best_model.hdf5\")\n",
    "results = model.predict(testGene, 180, verbose=1)\n",
    "save_result(r\"D:\\unet_ours\\data\\membrane\\test\", results)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "test1",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}