Evaluate horizontal and vertical padding invariance

twitter_saliency_filter_analysis--12.py

pairwise_tests_data = f'{img_dir}/Twitter_saliency/FairFace_pairwise_tests.csv'
if os.path.exists(pairwise_tests_data):
  pairwise_df = pd.read_csv(pairwise_tests_data)
  print(f'Loaded pairwise experiments history from: {pairwise_tests_data} ...')
  experiment_ids = list(pairwise_df['experiment_id'].values)
  instance_ids   = list(pairwise_df['instance_id'].values)
  img1           = list(pairwise_df['img1'].values)
  img2           = list(pairwise_df['img2'].values)
  baseline_h1    = list(pairwise_df['baseline_h1'].values)
  baseline_h2    = list(pairwise_df['baseline_h2'].values)
  baseline_v1    = list(pairwise_df['baseline_v1'].values)
  baseline_v2    = list(pairwise_df['baseline_v2'].values)
  saliency_out   = list(pairwise_df['saliency_out'].values)
  combine_mode   = list(pairwise_df['combine_mode'].values)
else:
  pairwise_df = pd.DataFrame()
  experiment_ids = []
  instance_ids   = []
  img1           = []
  img2           = []
  baseline_h1    = []
  baseline_h2    = []
  baseline_v1    = []
  baseline_v2    = []
  saliency_out   = []
  combine_mode   = []
  
debug = False

padding_eval = {'horizontal': {'padding_blocks': {1: {'max': 25, 'min': 0}}},
                'vertical': {'padding_blocks': {1: {'max': 25, 'min': 0}}}} if debug else \
              {'horizontal': { 
                    'padding_blocks': {
                         1: {'min': 0, 'max': 25},
                         2: {'min': 25, 'max': 75},
                         3: {'min': 75, 'max': 300},
                      }
                    },
                 'vertical': { 
                     'padding_blocks': {
                         1: {'min': 0, 'max': 25},
                         2: {'min': 25, 'max': 75},
                         3: {'min': 75, 'max': 300},
                     }
                   }
               }

output_dir =f'{img_dir}/Twitter_saliency/FairFace_pairwise_tests/'
num_eval = 1
for i in range(len(padding_eval)):
  eval_key = list(padding_eval.keys())[i]
  label_id = eval_key
  if  eval_key == 'horizontal':
    label_id = 'h'
    num_cols = 2
  elif eval_key == 'vertical':
    label_id = 'v'
    num_cols = 1
  padding_blocks = padding_eval[eval_key]['padding_blocks'] 
  for j in range(len(padding_blocks)):
    for k in tqdm(range(num_eval)):
      instance_id = randomID_generator()
      image_files = glob.glob(str(data_dir / Path("./*.jpg")))
      random.SystemRandom().shuffle(image_files)
      images = [Image.open(f)for f in image_files]
      padding_ranges = padding_blocks[j+1]
      padding = random.SystemRandom().choice(range(padding_ranges['min'],
                                                   padding_ranges['max']))
      print(f'Using a padding value: {padding}')
      img = join_images(images, col_wrap=num_cols, img_size=(128,128),
                        padding=padding)
      filename = f'{instance_id}_{file_id}_p{padding}_t{k}_{label_id}'
      output_file = f"{output_dir}/{filename}.jpeg"
      img.save(output_file, "JPEG")
      model.plot_img_crops_using_img(img, topK=5, col_wrap=6)
      saliency_info,sp = saliency_point_to_info(output_file, img_files, model, img_labels, image_mode='horizontal')
      plt.savefig(f"{output_dir}/{filename}_sm.jpeg",bbox_inches="tight")
      
model.plot_img_crops(data_dir / Path(f"{img_id1}.jpg"), topK=2, aspectRatios=[0.56])
plt.savefig(f"{img_id1}_{img_race1}_{img_gender1}_saliency.jpeg", bbox_inches="tight")

model.plot_img_crops(data_dir / Path(f"{img_id2}.jpg"), topK=2, aspectRatios=[0.56])
plt.savefig(f"{img_id2}_{img_race2}_{img_gender2}_saliency.jpeg", bbox_inches="tight")