My final thesis
Project coded in Python 3.6.4
Go to RUN directory.
Run following for installing necessary packages.
pip install -r requirements.txt
After installing packages simply run python script mitosis_gui.py
python3 mitosis_gui.py
For training new dataset with CNN model:
model = MITOSIS_CNN(IMG_SIZE, weights=weights_path, channels=10)
model.generate_train_data(MITOSIS_PATH,NONMITOSIS_PATH)
model.train_model(epochs=30,min_delta=0.01,cross_validation=0.2)
####################### Explanation of files in 'Code2 directory: avg.npy -> discriminative vector for generating DI image DI_generate.py -> generates DI image from multispectral images mitos_model_v2.py -> contains CNN model for predicting also for training model_v2_test.py -> test all images in ICPR dataset, also counts true/false positive/negative rate save_patterns_with_mitosis_M.py -> generate train data stacking all bands in numpy dump file save_patterns_with_nonmitosis_M.py -> generate train data stacking all bands in numpy dump file weights_multispectral_v2 -> pretrained CNN model