testing.py

"""
Copyright (c) 2025 Samsung Electronics Co., Ltd.

Author(s):
Mahmoud Afifi (m.afifi1@samsung.com, m.3afifi@gmail.com)

Licensed under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) License, (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at https://creativecommons.org/licenses/by-nc/4.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an
"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and limitations under the License.
For conditions of distribution and use, see the accompanying LICENSE.md file.

Testing script.
"""

from torch.utils.data import DataLoader
import argparse

import utils
from model import IllumEstimator
from dataloader import IlluminantEstimationDataLoader
from utils import *


def get_args():
  parser = argparse.ArgumentParser(
    description='Script for testing a trained model.'
  )
  parser.add_argument('--dataset_path', type=str, required=True,
                      help='Path to the dataset directory. The directory must contain subdirectories:'
                           ' "test" and "val".')
  parser.add_argument('--set_name', type=str, required=True, choices=TESTING_SET_NAMES,
                      help='Set name. Options: "test" and "val".')
  parser.add_argument('--model_path', type=str, default='./models',
                      help='Path to the directory of trained models.')
  parser.add_argument('--output_path', type=str, default='./results',
                      help='Path to the output directory for saving the results.')
  parser.add_argument('--device', type=str, choices=DEVICES, default='gpu',
                      help='Device to use for training. Options: "cpu" or "gpu".')
  parser.add_argument('--without_mask', action='store_true',
                      help='To use testing/validation images without mask.')
  parser.add_argument('--exp_name', type=str, default='awb',
                      help='Experiment name.')
  return parser.parse_args()

def testing(model: IllumEstimator, data: DataLoader):
  """Testing trained model."""
  model.eval()
  filenames = []
  estimated_illums = []
  for i, batch in enumerate(data):
    capture_data = batch['capture_data'].to(device=device)
    hist = batch['hist'].to(device=device)
    filenames.append(batch['filename'])
    est_illum = model(hist, capture_data).detach().cpu().numpy().tolist()
    estimated_illums.append(est_illum)
  return estimated_illums, filenames

if __name__ == '__main__':
  args = get_args()
  if args.set_name == 'val':
    valid = True
    test = False
  elif args.set_name == 'test':
    valid = False
    test = True
  else:
    raise ValueError('Invalid set_name.')

  if args.device == 'gpu':
    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
  else:
    device = torch.device('cpu')
  configs = utils.read_json_file(os.path.join(args.model_path, f'config-{args.exp_name}.json'))
  capture_data = configs['capture_data']
  user_preference = configs['user_preference']
  if user_preference:
    sub_folder = 'pref_illum'
  else:
    sub_folder = 'gt_illum'
  norm_values = configs['norm_values']
  hist_boundaries = configs['hist_boundaries']
  hist_bins = configs['hist_bins']
  target_size = configs['target_size']

  test_dataset = IlluminantEstimationDataLoader(
    data_dir=args.dataset_path, capture_data=capture_data, target_size=target_size,
    valid=valid, test=test, normalization_data=norm_values, hist_bins=hist_bins,
    hist_boundaries=hist_boundaries, without_mask=args.without_mask)

  test_loader = DataLoader(test_dataset, batch_size=1, shuffle=False, num_workers=NUM_WORKERS)

  capture_data_size = 0
  for i, data in enumerate(test_loader):
    capture_data_size = data['capture_data'][0, ...].numel()
    if i == 0:
      break

  model = IllumEstimator(in_channels=capture_data_size, hist_channels=4).to(device=device)
  folder = 'without_mask' if args.without_mask else 'with_mask'

  model_path = os.path.join(args.model_path, f'model-{args.exp_name}.pt')
  model.load_state_dict(torch.load(model_path, weights_only=True))
  model.print_num_of_params()
  estimated_illums, filenames = testing(model, test_loader)
  output_path = os.path.join(args.output_path, folder, sub_folder, f'{args.exp_name}-{args.set_name}')
  os.makedirs(output_path, exist_ok=True)
  for f, est_illum in zip(filenames, estimated_illums):
    utils.write_json_file({'est_illum': est_illum[0]}, os.path.join(output_path, f[0]))
  print('Done!')