Last tested on tensorflow 1.13.1
source : https://github.com/qqwweee/keras-yolo3
package:
- keras
- tensorflow
- pillow
- matplotlib
- numpy
- opencv
- kito
`python convert.py -w yolov3.cfg yolov3.weights model_data/yolo_weights.h5`
`python convert.py yolov3.cfg yolov3.weights model_data/yolo.h5`
For VOC dataset, try python voc_annotation.py
python yolo_video.py --input test_data/akiha.mp4 --model trained_yolo.h5 --anchors anchors/yolo_anchors.txt --classes class/voc_classes.txt
python yolo_video.py --input test_data/akiha.mp4 --model trained_yolo.h5 --anchors anchors/yolo_anchors.txt --classes class/voc_classes.txt
python mobilenet_video.py --input test_data/akiha.mp4
python train.py
python train.py --classes class_file test_data/akiha.mp4 -model model_file --anchors anchor_file
python evaluate.py -c eval_config.json
tensorboard --logdir=logs/000 --port=6007
train batch size yolo -> 18,12 small model -> 64 , 24
small mobilenet yolo for now {0: 0.2886880500915588, 1: 0.49941118603127055, 2: 0.2444640965334233, 3: 0.16006168349920366, 4: 0.1948608902713586, 5: 0.5300083075861101, 6: 0.5409369836908366, 7: 0.5067628271579261, 8: 0.10056515957446807, 9: 0.3287429704172446, 10: 0.23961672473867596, 11: 0.41025285247998355, 12: 0.47350048872418604, 13: 0.4251144456578134, 14: 0.42984661736129237, 15: 0.09929483514389173, 16: 0.23318323608432168, 17: 0.3794570650635555, 18: 0.4837190072005488, 19: 0.3191247696198191} aeroplane: 0.2887 bicycle: 0.4994 bird: 0.2445 boat: 0.1601 bottle: 0.1949 bus: 0.5300 car: 0.5409 cat: 0.5068 chair: 0.1006 cow: 0.3287 diningtable: 0.2396 dog: 0.4103 horse: 0.4735 motorbike: 0.4251 person: 0.4298 pottedplant: 0.0993 sheep: 0.2332 sofa: 0.3795 train: 0.4837 tvmonitor: 0.3191 mAP: 0.3444
small mobilenet first loss : ep003-loss113.347-val_loss94.183 middle loss : ep027-loss19.193-val_loss21.659.h5 last loss : last_loss14.0924-val_loss14.0924
Yolo aeroplane: 0.7073 bicycle: 0.7195 bird: 0.5802 boat: 0.4040 bottle: 0.5104 bus: 0.7690 car: 0.7655 cat: 0.7586 chair: 0.4294 cow: 0.6622 diningtable: 0.5250 dog: 0.7139 horse: 0.7381 motorbike: 0.6654 person: 0.6838 pottedplant: 0.2941 sheep: 0.5743 sofa: 0.6178 train: 0.7244 tvmonitor: 0.5933 mAP: 0.6218