mrq/DL-Art-School
1
3
Fork 5
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Allow test to operate on batches

Browse Source
This commit is contained in:
James Betker 2020-04-23 23:59:09 -06:00
parent 8ead9ae183
commit e98d92fc77
2 changed files with 92 additions and 83 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
codes/data/__init__.py
View File

@ -21,7 +21,7 @@ def create_dataloader(dataset, dataset_opt, opt=None, sampler=None):
num_workers=num_workers, sampler=sampler, drop_last=True, num_workers=num_workers, sampler=sampler, drop_last=True,
pin_memory=False) pin_memory=False)
else: else:
return torch.utils.data.DataLoader(dataset, batch_size=1, shuffle=False, num_workers=0, return torch.utils.data.DataLoader(dataset, batch_size=12, shuffle=False, num_workers=3,
pin_memory=False) pin_memory=False)
@ -32,8 +32,9 @@ def create_dataset(dataset_opt):
from data.LQ_dataset import LQDataset as D from data.LQ_dataset import LQDataset as D
elif mode == 'LQGT': elif mode == 'LQGT':
from data.LQGT_dataset import LQGTDataset as D from data.LQGT_dataset import LQGTDataset as D
elif mode == 'GTLQ': # datasets for image corruption
from data.GTLQ_dataset import GTLQDataset as D elif mode == 'downsample':
from data.Downsample_dataset import DownsampleDataset as D
# datasets for video restoration # datasets for video restoration
elif mode == 'REDS': elif mode == 'REDS':
from data.REDS_dataset import REDSDataset as D from data.REDS_dataset import REDSDataset as D

168
codes/test.py
View File

@ -9,97 +9,105 @@ import utils.util as util
from data.util import bgr2ycbcr from data.util import bgr2ycbcr
from data import create_dataset, create_dataloader from data import create_dataset, create_dataloader
from models import create_model from models import create_model
from tqdm import tqdm
#### options if __name__ == "__main__":
parser = argparse.ArgumentParser() #### options
parser.add_argument('-opt', type=str, help='Path to options YMAL file.', default='options/test/test_ESRGAN_vrp.yml') want_just_images = True
opt = option.parse(parser.parse_args().opt, is_train=False) parser = argparse.ArgumentParser()
opt = option.dict_to_nonedict(opt) parser.add_argument('-opt', type=str, help='Path to options YMAL file.', default='options/test/test_corrupt_vixen_adrianna.yml')
opt = option.parse(parser.parse_args().opt, is_train=False)
opt = option.dict_to_nonedict(opt)
util.mkdirs( util.mkdirs(
(path for key, path in opt['path'].items() (path for key, path in opt['path'].items()
if not key == 'experiments_root' and 'pretrain_model' not in key and 'resume' not in key)) if not key == 'experiments_root' and 'pretrain_model' not in key and 'resume' not in key))
util.setup_logger('base', opt['path']['log'], 'test_' + opt['name'], level=logging.INFO, util.setup_logger('base', opt['path']['log'], 'test_' + opt['name'], level=logging.INFO,
screen=True, tofile=True) screen=True, tofile=True)
logger = logging.getLogger('base') logger = logging.getLogger('base')
logger.info(option.dict2str(opt)) logger.info(option.dict2str(opt))
#### Create test dataset and dataloader #### Create test dataset and dataloader
test_loaders = [] test_loaders = []
for phase, dataset_opt in sorted(opt['datasets'].items()): for phase, dataset_opt in sorted(opt['datasets'].items()):
test_set = create_dataset(dataset_opt) test_set = create_dataset(dataset_opt)
test_loader = create_dataloader(test_set, dataset_opt) test_loader = create_dataloader(test_set, dataset_opt)
logger.info('Number of test images in [{:s}]: {:d}'.format(dataset_opt['name'], len(test_set))) logger.info('Number of test images in [{:s}]: {:d}'.format(dataset_opt['name'], len(test_set)))
test_loaders.append(test_loader) test_loaders.append(test_loader)
model = create_model(opt) model = create_model(opt)
for test_loader in test_loaders: for test_loader in test_loaders:
test_set_name = test_loader.dataset.opt['name'] test_set_name = test_loader.dataset.opt['name']
logger.info('\nTesting [{:s}]...'.format(test_set_name)) logger.info('\nTesting [{:s}]...'.format(test_set_name))
test_start_time = time.time() test_start_time = time.time()
dataset_dir = osp.join(opt['path']['results_root'], test_set_name) dataset_dir = osp.join(opt['path']['results_root'], test_set_name)
util.mkdir(dataset_dir) util.mkdir(dataset_dir)
test_results = OrderedDict() test_results = OrderedDict()
test_results['psnr'] = [] test_results['psnr'] = []
test_results['ssim'] = [] test_results['ssim'] = []
test_results['psnr_y'] = [] test_results['psnr_y'] = []
test_results['ssim_y'] = [] test_results['ssim_y'] = []
for data in test_loader: tq = tqdm(test_loader)
need_GT = False if test_loader.dataset.opt['dataroot_GT'] is None else True for data in tq:
model.feed_data(data, need_GT=need_GT) need_GT = False if test_loader.dataset.opt['dataroot_GT'] is None else True
img_path = data['GT_path'][0] if need_GT else data['LQ_path'][0] model.feed_data(data, need_GT=need_GT)
img_name = osp.splitext(osp.basename(img_path))[0] model.test()
model.test() visuals = model.fake_H.detach().float().cpu()
visuals = model.get_current_visuals(need_GT=need_GT) for i in range(visuals.shape[0]):
img_path = data['GT_path'][i] if need_GT else data['LQ_path'][i]
img_name = osp.splitext(osp.basename(img_path))[0]
sr_img = util.tensor2img(visuals['rlt']) # uint8 sr_img = util.tensor2img(visuals[i]) # uint8
# save images # save images
suffix = opt['suffix'] suffix = opt['suffix']
if suffix: if suffix:
save_img_path = osp.join(dataset_dir, img_name + suffix + '.png') save_img_path = osp.join(dataset_dir, img_name + suffix + '.png')
else: else:
save_img_path = osp.join(dataset_dir, img_name + '.png') save_img_path = osp.join(dataset_dir, img_name + '.png')
util.save_img(sr_img, save_img_path) util.save_img(sr_img, save_img_path)
# calculate PSNR and SSIM if want_just_images:
if need_GT: continue
gt_img = util.tensor2img(visuals['GT'])
sr_img, gt_img = util.crop_border([sr_img, gt_img], opt['scale'])
psnr = util.calculate_psnr(sr_img, gt_img)
ssim = util.calculate_ssim(sr_img, gt_img)
test_results['psnr'].append(psnr)
test_results['ssim'].append(ssim)
if gt_img.shape[2] == 3: # RGB image # calculate PSNR and SSIM
sr_img_y = bgr2ycbcr(sr_img / 255., only_y=True) if need_GT:
gt_img_y = bgr2ycbcr(gt_img / 255., only_y=True) gt_img = util.tensor2img(visuals['GT'])
sr_img, gt_img = util.crop_border([sr_img, gt_img], opt['scale'])
psnr = util.calculate_psnr(sr_img, gt_img)
ssim = util.calculate_ssim(sr_img, gt_img)
test_results['psnr'].append(psnr)
test_results['ssim'].append(ssim)
psnr_y = util.calculate_psnr(sr_img_y * 255, gt_img_y * 255) if gt_img.shape[2] == 3: # RGB image
ssim_y = util.calculate_ssim(sr_img_y * 255, gt_img_y * 255) sr_img_y = bgr2ycbcr(sr_img / 255., only_y=True)
test_results['psnr_y'].append(psnr_y) gt_img_y = bgr2ycbcr(gt_img / 255., only_y=True)
test_results['ssim_y'].append(ssim_y)
logger.info(
'{:20s} - PSNR: {:.6f} dB; SSIM: {:.6f}; PSNR_Y: {:.6f} dB; SSIM_Y: {:.6f}.'.
format(img_name, psnr, ssim, psnr_y, ssim_y))
else:
logger.info('{:20s} - PSNR: {:.6f} dB; SSIM: {:.6f}.'.format(img_name, psnr, ssim))
else:
logger.info(img_name)
if need_GT: # metrics psnr_y = util.calculate_psnr(sr_img_y * 255, gt_img_y * 255)
# Average PSNR/SSIM results ssim_y = util.calculate_ssim(sr_img_y * 255, gt_img_y * 255)
ave_psnr = sum(test_results['psnr']) / len(test_results['psnr']) test_results['psnr_y'].append(psnr_y)
ave_ssim = sum(test_results['ssim']) / len(test_results['ssim']) test_results['ssim_y'].append(ssim_y)
logger.info( logger.info(
'----Average PSNR/SSIM results for {}----\n\tPSNR: {:.6f} dB; SSIM: {:.6f}\n'.format( '{:20s} - PSNR: {:.6f} dB; SSIM: {:.6f}; PSNR_Y: {:.6f} dB; SSIM_Y: {:.6f}.'.
test_set_name, ave_psnr, ave_ssim)) format(img_name, psnr, ssim, psnr_y, ssim_y))
if test_results['psnr_y'] and test_results['ssim_y']: else:
ave_psnr_y = sum(test_results['psnr_y']) / len(test_results['psnr_y']) logger.info('{:20s} - PSNR: {:.6f} dB; SSIM: {:.6f}.'.format(img_name, psnr, ssim))
ave_ssim_y = sum(test_results['ssim_y']) / len(test_results['ssim_y']) else:
logger.info(img_name)
if not want_just_images and need_GT: # metrics
# Average PSNR/SSIM results
ave_psnr = sum(test_results['psnr']) / len(test_results['psnr'])
ave_ssim = sum(test_results['ssim']) / len(test_results['ssim'])
logger.info( logger.info(
'----Y channel, average PSNR/SSIM----\n\tPSNR_Y: {:.6f} dB; SSIM_Y: {:.6f}\n'. '----Average PSNR/SSIM results for {}----\n\tPSNR: {:.6f} dB; SSIM: {:.6f}\n'.format(
format(ave_psnr_y, ave_ssim_y)) test_set_name, ave_psnr, ave_ssim))
if test_results['psnr_y'] and test_results['ssim_y']:
ave_psnr_y = sum(test_results['psnr_y']) / len(test_results['psnr_y'])
ave_ssim_y = sum(test_results['ssim_y']) / len(test_results['ssim_y'])
logger.info(
'----Y channel, average PSNR/SSIM----\n\tPSNR_Y: {:.6f} dB; SSIM_Y: {:.6f}\n'.
format(ave_psnr_y, ave_ssim_y))