| import os |
| import numpy as np |
|
|
| import torch |
| from tqdm import tqdm |
| from vbench.utils import load_video, load_dimension_info, clip_transform, read_frames_decord_by_fps, CACHE_DIR |
| from vbench.third_party.ViCLIP.viclip import ViCLIP |
| from vbench.third_party.ViCLIP.simple_tokenizer import SimpleTokenizer |
|
|
| def get_text_features(model, input_text, tokenizer, text_feature_dict={}): |
| if input_text in text_feature_dict: |
| return text_feature_dict[input_text] |
| text_template= f"{input_text}" |
| with torch.no_grad(): |
| text_features = model.encode_text(text_template).float() |
| text_features /= text_features.norm(dim=-1, keepdim=True) |
| text_feature_dict[input_text] = text_features |
| return text_features |
|
|
| def get_vid_features(model, input_frames): |
| with torch.no_grad(): |
| clip_feat = model.encode_vision(input_frames,test=True).float() |
| clip_feat /= clip_feat.norm(dim=-1, keepdim=True) |
| return clip_feat |
|
|
| def get_predict_label(clip_feature, text_feats_tensor, top=5): |
| label_probs = (100.0 * clip_feature @ text_feats_tensor.T).softmax(dim=-1) |
| top_probs, top_labels = label_probs.cpu().topk(top, dim=-1) |
| return top_probs, top_labels |
|
|
| def overall_consistency(clip_model, video_pairs, tokenizer, device, sample="middle"): |
| sim = [] |
| video_results = [] |
| image_transform = clip_transform(224) |
| for info in tqdm(video_pairs): |
| query = info['prompt'] |
| |
| video_path = info['content_path'] |
|
|
| with torch.no_grad(): |
| images = read_frames_decord_by_fps(video_path, num_frames=8, sample=sample) |
| images = image_transform(images) |
| images = images.to(device) |
| clip_feat = get_vid_features(clip_model,images.unsqueeze(0)) |
| text_feat = get_text_features(clip_model, query, tokenizer) |
| logit_per_text = clip_feat @ text_feat.T |
| score_per_video = float(logit_per_text[0][0].cpu()) |
| sim.append(score_per_video) |
| video_results.append({'prompt':query, 'video_path': video_path, 'video_results': score_per_video}) |
| avg_score = np.mean(sim) |
| |
| return { |
| "score":[avg_score, video_results] |
| } |
|
|
|
|
| def compute_overall_consistency(video_pairs): |
| device = torch.device("cuda") |
| submodules_list = {"pretrain": f'{CACHE_DIR}/ViCLIP/ViClip-InternVid-10M-FLT.pth'} |
| tokenizer = SimpleTokenizer(os.path.join(CACHE_DIR, "ViCLIP/bpe_simple_vocab_16e6.txt.gz")) |
| viclip = ViCLIP(tokenizer= tokenizer, **submodules_list).to(device) |
| results = overall_consistency(viclip, video_pairs, tokenizer, device) |
| return results |
|
|
|
|
|
|
