Brain_cancer_detection_on_Raspberry_Pi

@@ First check whether git is installed in Rpi3b+@@

! git --version

If this fails with a error message install git ...

! sudo apt install git

@@ Clone this repository@@

! git clone https://github.com/Soham-coder/Raspberry_pi_edge_ML.git

@@ Run normal_quantized code@@
Go inside directory Normal_quantized

! sudo chmod +x script_normal.sh

! ./script_normal.sh

It will give print "Press pushbutton for predicting images"

Press pushbutton connected to GPIO-15

It will ask for prompt "Enter path of image directory where images are present:"

Give prompt input as ./test_orig

- Note: Don't press pushbutton until current prediction operation is finished, else it will ask for next prompt

Once all finish, press ctrl+c

Output log - ./log/out.txt

Normal_quantized_log

+ Average Prediction Time: 5.3895740906397505 Seconds
+ Average Size of test images - 24035 bytes
+ Pre-trained Model Size - 7255312 bytes
+ Accuracy - 97.92% 

@@ Run dwt_quantized code@@
Go inside directory DWT_quantized

! sudo chmod +x script_dwt.sh

! ./script_dwt.sh

It will give print "Press pushbutton for predicting images"

Press pushbutton connected to GPIO-15

It will ask for prompt "Enter path of image directory where images are present:"

Give prompt input as ./test_dwt

- Note: Don't press pushbutton until current prediction operation is finished, else it will ask for next prompt

Once all finish, press ctrl+c

Output log - ./log/out.txt

DWT_quantized_log

+ Average Prediction Time: 0.24476244866847993 Seconds
+ Average Size of test images - 5894 bytes
+ Pre-trained Model Size - 795552 bytes
+ Accuracy - 95.33%

@@ Run vgg16_quantized code@@
Go inside directory VGG16_quantized

! sudo chmod +x script_vgg16.sh

! ./script_vgg16.sh

It will give print "Press pushbutton for predicting images"

Press pushbutton connected to GPIO-15

It will ask for prompt "Enter path of image directory where images are present:"

Give prompt input as ./test_orig

- Note: Don't press pushbutton until current prediction operation is finished, else it will ask for next prompt

Once all finish, press ctrl+c

Output log - ./log/out.txt

VGG16_quantized

+ Average Prediction Time: 30.64683867924249 Seconds
+ Average Size of test images - 24035 bytes
+ Pre-trained Model Size - 18143072 bytes

Accuracy_logs

Normal_accuracy

DWT_accuracy

@@ Run Accuracy scripts@@
Go inside directory Calculate_accuracy
@@ Run normal accuracy@@
! sudo chmod +x script_accuracy_normal.sh
! ./script_accuracy_normal.sh
Output logs - 
# ./log/normal.txt

@@ Run dwt accuracy@@
! sudo chmod +x script_accuracy_dwt.sh
! ./script_accuracy_dwt.sh
Output logs -
# ./log/dwt.txt

@@ Run cs accuracy@@
! sudo chmod +x script_accuracy_cs.sh
! ./script_accuracy_cs.sh
Output logs - 
# ./log/cs.txt

Final_Report

Final Report

Utility

### Utility scripts

average_prediction_runtime.py - Parses the log file and calculates average prediction time
find_avg_test_image_size.sh - Calculates the average image size of test images in test image folder in bytes
find_model_size.sh - Calculates the Pretrained model size in weights folder
install_git.sh - Incase you want to install git in Linux-Ubuntu based system through apt package manager, use this
form_dict.py - Creates Dictionary/associative array of "image name" & "category"- Enter prompt as ./test_dwt

install_tf_2_4_0_rpi3b.sh - Install tf==2.4.0 in Rpi 3b+/4 in Debian Raspbian OS
cd Install_tf-2.4.0-armv71-rpi3b
chmod +x install_tf_2_4_0_rpi3b.sh
./install_tf_2_4_0_rpi3b.sh
[Restart the Terminal](needed)

P.S - tf=2.4.0 should get installed in pip3, python3

Project Maintainers

Arindam Majee

Soham Mondal