Computer Vision News - November 2022

37 Glaucoma and Segmentation Here we create the train and test variables. The example shown here it’s just for variable one but similar would be for variable two. I will leave this here as a small exercise for the reader! rng = np.random.RandomState(rand_sv) #DRISHTI X_train1t, X_test1t, Y_train1t, Y_test1t, F_train1t, F_test1t ,ind_train1, ind_test1= train_test_split( X, Yb, FC, indDRI,test_size=0.25,random_state=rng) #RIM ONE X_train2, X_test2, Y_train2, Y_test2, F_train2, F_test2, ind_train2, ind_test2= train_ test_split( Xv, Yvf, FCv, indRIM, test_size=0.25,random_state=rng) X_test_dri=np.copy(X_test1t) Y_test_dri=np.copy(Y_test1t) F_test_dri=np.copy(F_test1t) Preprocessing for cup segmentation Here for each image CLAHE is performed like this: for i in range(16): X_traine[i*size:(i+1)*size]=X_train1 Y_traine[i*size:(i+1)*size]=Y_train1 base= 16*size for j in tqdm_notebook(range(X_train1.shape[0])): X_traine[base]=skimage.exposure.equalize_adapthist(X_train1[j], clip_limit=0.04) Y_traine[base]=Y_train1[j] base+=1 X_traine[base]=skimage.exposure.equalize_adapthist(X_train1[j], clip_limit=0.02) Y_traine[base]=Y_train1[j] base+=1 X_traine[base]=modify_brightness_p(X_train1[j],0.9) Y_traine[base]=Y_train1[j] base+=1 X_traine[base]=modify_brightness_p(X_train1[j],1.1) Y_traine[base]=Y_train1[j] base+=1 X_testc=np.copy(X_test) Y_testc=np.copy(Y_test)