To extract frames from the video, use jon_data_2_frames.py (check_dirs.py)
To obtain frames in use: label_useful_frames.py (config.py, check_dirs.py)
To generate cropped image frames, ground truth heatmap(cropped), gaze log files(cropped), frames with plotted gaze points and a video for visualization, please use validation.py (config.py, video.py, heatmap.py, check_dirs.py, randomly_crop_image.py, read_frameInfo.py, resize_patch_for_cnn.py)
To flip img/heatmap/ground truth: flip_img.py(check_dirs.py, config.py)
Base line model(Itti, chance, anti_saliency): main.py (config.py, check_dirs.py, itti_model.py, heatmap.py, metrics.py, pySaliencyMap.py, pySaliencyMapDefs.py)
DNN models: VGG.py (util_calculations.py, load_data_for_cnn.py, base_misc_utils.py)
To replay predictions: replay_video.py (check_dirs.py)