valid_pixels_labels = [] valid_superpixels = [] for i in xrange(0,num_images): feature = Feature() feature.create(image_filenames[i],label_image_filenames[i]) labels = feature.getSuperpixelLabels() feature_vectors = feature.getFeaturesVectors() h, img_filename = os.path.split(image_filenames[i]) h, label_img_filename = os.path.split(label_image_filenames[i]) folder = "extracted/" + img_filename + "-" + label_img_filename + "/" if not os.path.exists(folder): os.makedirs(folder) scipy.io.savemat(folder+"superpixel", {'features':feature.getSuperpixelImage()}, oned_as='column') scipy.io.savemat(folder+"superpixel_labels", {'features':feature.getSuperpixelLabels()}, oned_as='column') scipy.io.savemat(folder+"location", {'features':feature.getSuperpixelsLocation()}, oned_as='column') scipy.io.savemat(folder+"color", {'features':feature.getSuperpixelsColor()}, oned_as='column') scipy.io.savemat(folder+"hog", {'features':feature.getSuperpixelsHog()}, oned_as='column') scipy.io.savemat(folder+"size", {'features':feature.getSuperpixelsSize()}, oned_as='column') scipy.io.savemat(folder+"texture", {'features':feature.getSuperpixelsTexture()}, oned_as='column') scipy.io.savemat(folder+"features_combined", {'features':feature.getFeaturesVectors()}, oned_as='column') if file_labels[i] != TESTING: # store data if file_labels[i] == TRAINING: train_superpixels.append(feature.getSuperpixelImage()) train_labels = np.append(train_labels, labels, 0) if train_data==[]: train_data = feature_vectors