| import torch |
| from tqdm import tqdm |
| from torchvision import transforms |
| from pyiqa.archs.musiq_arch import MUSIQ |
| from vbench.utils import load_video, load_dimension_info |
|
|
| from .distributed import ( |
| get_world_size, |
| get_rank, |
| all_gather, |
| barrier, |
| distribute_list_to_rank, |
| gather_list_of_dict, |
| ) |
|
|
|
|
| def transform(images, preprocess_mode='shorter'): |
| if preprocess_mode.startswith('shorter'): |
| _, _, h, w = images.size() |
| if min(h,w) > 512: |
| scale = 512./min(h,w) |
| images = transforms.Resize(size=( int(scale * h), int(scale * w) ), antialias=False)(images) |
| if preprocess_mode == 'shorter_centercrop': |
| images = transforms.CenterCrop(512)(images) |
|
|
| elif preprocess_mode == 'longer': |
| _, _, h, w = images.size() |
| if max(h,w) > 512: |
| scale = 512./max(h,w) |
| images = transforms.Resize(size=( int(scale * h), int(scale * w) ), antialias=False)(images) |
|
|
| elif preprocess_mode == 'None': |
| return images / 255. |
|
|
| else: |
| raise ValueError("Please recheck imaging_quality_mode") |
| return images / 255. |
|
|
| def technical_quality(model, video_list, device, **kwargs): |
| if 'imaging_quality_preprocessing_mode' not in kwargs: |
| preprocess_mode = 'longer' |
| else: |
| preprocess_mode = kwargs['imaging_quality_preprocessing_mode'] |
| video_results = [] |
| for video_path in tqdm(video_list, disable=get_rank() > 0): |
| images = load_video(video_path) |
| images = transform(images, preprocess_mode) |
| acc_score_video = 0. |
| for i in range(len(images)): |
| frame = images[i].unsqueeze(0).to(device) |
| score = model(frame) |
| acc_score_video += float(score) |
| video_results.append({'video_path': video_path, 'video_results': acc_score_video/len(images)}) |
| average_score = sum([o['video_results'] for o in video_results]) / len(video_results) |
| average_score = average_score / 100. |
| return average_score, video_results |
|
|
|
|
| def compute_imaging_quality(json_dir, device, submodules_list, **kwargs): |
| model_path = submodules_list['model_path'] |
|
|
| model = MUSIQ(pretrained_model_path=model_path) |
| model.to(device) |
| model.training = False |
| |
| video_list, _ = load_dimension_info(json_dir, dimension='imaging_quality', lang='en') |
| video_list = distribute_list_to_rank(video_list) |
| all_results, video_results = technical_quality(model, video_list, device, **kwargs) |
| if get_world_size() > 1: |
| video_results = gather_list_of_dict(video_results) |
| all_results = sum([d['video_results'] for d in video_results]) / len(video_results) |
| all_results = all_results / 100. |
| return all_results, video_results |
|
|
