| |
| from llava.model.builder import load_pretrained_model |
| from llava.mm_utils import get_model_name_from_path, process_images, tokenizer_image_token |
| from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN, IGNORE_INDEX |
| from llava.conversation import conv_templates, SeparatorStyle |
| from PIL import Image |
| import requests |
| import copy |
| import torch |
| import sys |
| import warnings |
| from decord import VideoReader, cpu |
| import numpy as np |
| import json |
| import argparse |
| import os |
| from vbench2.utils import load_dimension_info |
| from tqdm import tqdm |
| warnings.filterwarnings("ignore") |
|
|
| def load_video(video_path, max_frames_num,fps=1,force_sample=False): |
| if max_frames_num == 0: |
| return np.zeros((1, 336, 336, 3)) |
| vr = VideoReader(video_path, ctx=cpu(0),num_threads=1) |
| total_frame_num = len(vr) |
| video_time = total_frame_num / vr.get_avg_fps() |
| fps = round(vr.get_avg_fps()/fps) |
| frame_idx = [i for i in range(0, len(vr), fps)] |
| frame_time = [i/fps for i in frame_idx] |
| if len(frame_idx) > max_frames_num or force_sample: |
| sample_fps = max_frames_num |
| uniform_sampled_frames = np.linspace(0, total_frame_num - 1, sample_fps, dtype=int) |
| frame_idx = uniform_sampled_frames.tolist() |
| frame_time = [i/vr.get_avg_fps() for i in frame_idx] |
| frame_time = ",".join([f"{i:.2f}s" for i in frame_time]) |
| spare_frames = vr.get_batch(frame_idx).asnumpy() |
| return spare_frames,frame_time,video_time |
|
|
| def LLaVA_Video(prompt_dict_ls, model, tokenizer, image_processor, device): |
| final_score = 0 |
| valid_num = 0 |
| processed_json=[] |
| for prompt_dict in tqdm(prompt_dict_ls): |
| base_question = prompt_dict['auxiliary_info'] |
| question_num = len(base_question) |
| video_paths = prompt_dict['video_list'] |
| for video_path in video_paths: |
| max_frames_num = 64 |
| video,frame_time,video_time = load_video(video_path, max_frames_num, 1, force_sample=True) |
| video = image_processor.preprocess(video, return_tensors="pt")["pixel_values"].cuda().bfloat16() |
| video = [video] |
| conv_template = "qwen_1_5" |
| time_instruciton = f"The video lasts for {video_time:.2f} seconds, and {len(video[0])} frames are uniformly sampled from it. These frames are located at {frame_time}. Please answer yes or no only for the following questions." |
| flag=True |
| for i in range(question_num): |
| question = DEFAULT_IMAGE_TOKEN + f"{time_instruciton}\n{base_question[i]}" |
| conv = copy.deepcopy(conv_templates[conv_template]) |
| conv.append_message(conv.roles[0], question) |
| conv.append_message(conv.roles[1], None) |
| prompt_question = conv.get_prompt() |
| input_ids = tokenizer_image_token(prompt_question, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt").unsqueeze(0).to(device) |
| cont = model.generate( |
| input_ids, |
| images=video, |
| modalities= ["video"], |
| do_sample=False, |
| temperature=0, |
| max_new_tokens=4096, |
| ) |
| text_outputs = tokenizer.batch_decode(cont, skip_special_tokens=True)[0] |
| if "yes" not in text_outputs.lower(): |
| flag=False |
| if flag: |
| final_score+=1 |
| sco=1 |
| else: |
| sco=0 |
| valid_num+=1 |
| processed_json.append({'video_path': video_path, 'video_results': sco}) |
| return final_score/valid_num, processed_json |
| |
| |
| def compute_dynamic_attribute(json_dir, device, submodules_dict, **kwargs): |
| _, prompt_dict_ls = load_dimension_info(json_dir, dimension='dynamic_attribute', lang='en') |
| |
| model_name = "llava_qwen" |
| device_map = "auto" |
| try: |
| pretrained = submodules_dict['llava'] |
| llava_tokenizer, llava_model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, torch_dtype="bfloat16", device_map=device_map) |
| except: |
| pretrained = "lmms-lab/LLaVA-Video-7B-Qwen2" |
| llava_tokenizer, llava_model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, torch_dtype="bfloat16", device_map=device_map) |
| llava_model.eval() |
| |
| all_results, video_results = LLaVA_Video(prompt_dict_ls, llava_model, llava_tokenizer, image_processor, device) |
| all_results = sum([d['video_results'] for d in video_results]) / len(video_results) |
| return all_results, video_results |
