Computer Vision News - November 2022

36 Computer Vision Tool X=np.asarray(X) am=np.amax(X) X=X.astype(np.float32)/am print(np.amax(X),np.amin(X)) Y=[cv2.resize(img, (128, 128),interpolation=cv2.INTER_NEAREST)[..., None] for img in Yc] Y=np.asarray(Y) ym=np.amax(Y) Yf=Y/ym #get the binary mask Yb=(Y>0.5).astype(np.float32) print(np.amax(Yf),np.amin(Yf)) print(X.shape) 1.0 0.0 1.0 0.0 (50, 128, 128, 3) Similar to the RIM-ONE dataset. Xvori= h5f1['RIM-ONE v3/orig/images'] disc_locationsv = h5f1['RIM-ONE v3/512 px/disc_locations'] Glauv=np.asarray(h5f1['RIM-ONE v3/512 px/is_ill'] ) FCv=np.asarray(h5f1['RIM-ONE v3/512 px/file_codes']) indRIM=np.arange(0,Xvori.shape[0]) bsqsidev=np.maximum((disc_locationsv[:,3]-disc_locationsv[:,1]),(disc_locationsv[:,2]- disc_locationsv[:,0])) low_cont_n=0; ivsize=Xvori.shape[1] Yvf = h5f1['RIM-ONE v3/512 px/cup'] Xvc = [Xvori[i][index512_resize(disc_locationsv[i][0],ivsize):index512_ resize(disc_locationsv[i][0]+bsqsidev[i],ivsize), index512_resize(disc_locationsv[i] [1],ivsize):index512_resize(disc_locationsv[i][1]+bsqsidev[i],ivsize)] for i in range(len(Xvori))] Yvc=[Yvf[i][disc_locationsv[i][0]:disc_locationsv[i][0]+bsqsidev[i], disc_locationsv[i] [1]:disc_locationsv[i][1]+bsqsidev[i]] for i in range(len(Xvori))] Xvn=[cv2.resize(img, (128, 128),interpolation=cv2.INTER_NEAREST) for img in Xvc] Yv=[cv2.resize(img, (128, 128),interpolation=cv2.INTER_NEAREST)[..., None] for img in Yvc] Yv=np.asarray(Yv) ym=np.amax(Yv) Yvf=(Yv/ym).astype(np.float32) Train and test