Build pairwise image comparisons using the Twitter saliency filter

twitter_saliency_filter_analysis--09.py

clear_image_history(unzip_dir)
get_fairface_img(img_labels, img_idx1, unzip_dir, fairface_data)
get_fairface_img(img_labels, img_idx2, unzip_dir, fairface_data)

img_path = next(data_dir.glob("./*.jpg"))
print(img_path)

for img_file in data_dir.glob("./*.jpg"):
  img = mpimg.imread(img_file)
  plt.figure()
  plt.imshow(img)
  plt.gca().add_patch(
      Rectangle((0, 0), 200, 112, linewidth=1, edgecolor="r", facecolor="none")
  )
  
cmd = f"{str(bin_path)} {str(model_path)} '{img_path.absolute()}' show_all_points"
print(cmd)

output = subprocess.check_output(cmd, shell=True)  # Success!
print(output.splitlines())

!{str(bin_path)} {str(model_path)} '{img_path.absolute()}' show_all_points | head

parse_output(output).keys()

model = ImageSaliencyModel(crop_binary_path=bin_path, crop_model_path=model_path)

print(plt.matplotlib.__version__)
print(list(data_dir.glob("./*.jpg")))

for img_path in data_dir.glob("*.jpg"):
    print(img_path)
    model.plot_img_crops(img_path)
    
for img_path in reservoir_sampling(data_dir.glob("./*.jpg"), K=5):
  model.plot_img_crops(img_path)