ESTP-Bench/estp_dataset/benchmark/eval_cost.py

import json,os
import numpy as np

def load_multiple_json(file_path):
    """读取包含多个 JSON 对象的文件，并将每个 JSON 对象解析成 Python 对象，存放在列表中。"""
    with open(file_path, 'r', encoding='utf-8') as f:
        content = f.read()
    
    decoder = json.JSONDecoder()
    pos = 0
    results = []
    content_length = len(content)
    
    while pos < content_length:
        # 跳过空白字符
        while pos < content_length and content[pos].isspace():
            pos += 1
        if pos >= content_length:
            break
        try:
            obj, new_pos = decoder.raw_decode(content, pos)
            results.append(obj)
            pos = new_pos
        except json.JSONDecodeError as e:
            # 出现解析错误则退出循环
            print(f"JSON 解析错误: {e}")
            break
    return results



eval_file = '/root/videollm-online/data/estp_dataset/estpSqa_ours/LIVE_IT0.95.json'
eval_model = 'VideollmOnline' 

# eval_file = '/root/videollm-online/data/estp_dataset/estpSqa_ours_5cases/LivebaseStage2.json'
# eval_file = '/root/videollm-online/data/estp_dataset/estpSqa_ours/LivebaseStage3_high11_.json'
# eval_model = 'EWO'

# eval_file = '/root/videollm-online/data/estp_dataset/estpCqa_baseline/MiniCPMV_fbf_0.175.json'
# eval_model = 'MiniCPMV'

# eval_file = '/root/videollm-online/data/estp_dataset/estpSqa_baseline/LLaVAOneVision_passiveevaluator_deepseek_5_5.json'
# eval_model = 'LLaVAOneVision'

# eval_file = '/root/videollm-online/data/estp_dataset/estpSqa_baseline/LLaVANextVideo7B_passiveevaluator_deepseek_5_5.json'
# eval_model = 'LLaVANextVideo7B'

# eval_file = '/root/videollm-online/data/estp_dataset/estpSqa_baseline_5cases/Qwen2VL_fbf_5casesevaluator_deepseek_5_5.json'
# eval_model = 'Qwen2VL'

# eval_file = '/root/videollm-online/data/estp_dataset/estpSqa_baseline/InternVLV28_passiveevaluator_deepseek_5_5.json'
# eval_model = 'InternVLV28'


parent_dir = os.path.dirname(eval_file)
eval_files = [os.path.join(parent_dir, f) for f in os.listdir(parent_dir) if f.startswith(eval_file.split('/')[-1])]
eval_result = {}
for eval_file in eval_files:
    eval_result.update(json.load(open(eval_file)))

task2number = {
    "Object Recognition": 0,
    "Attribute Perception": 0,
    "Text-Rich Understanding": 0,
    "Object Localization": 0,
    "Object State Change Recognition": 0,
    "Ego Object Localization": 0,
    "Ego Object State Change Recognition": 0,
    "Action Recognition": 0,
    "Object Function": 0,
    "Information Function": 0,
    "Action Reasoning": 0,
    "Task Understanding": 0,
    
}

task2score = {
    "Object Recognition": 0,
    "Attribute Perception": 0,
    "Text-Rich Understanding": 0,
    "Object Localization": 0,
    "Object State Change Recognition": 0,
    "Ego Object Localization": 0,
    "Ego Object State Change Recognition": 0,
    "Action Recognition": 0,
    "Object Function": 0,
    "Information Function": 0,
    "Action Reasoning": 0,
    "Task Understanding": 0,
    
    
}

task2recall = {
    "Object Recognition": 0,
    "Attribute Perception": 0,
    "Text-Rich Understanding": 0,
    "Object Localization": 0,
    "Object State Change Recognition": 0,
    "Ego Object Localization": 0,
    "Ego Object State Change Recognition": 0,
    "Action Recognition": 0,
    "Object Function": 0,
    "Information Function": 0,
    "Action Reasoning": 0,
    "Task Understanding": 0,
    
    
}

task2recall_score = {
    "Object Recognition": 0,
    "Attribute Perception": 0,
    "Text-Rich Understanding": 0,
    "Object Localization": 0,
    "Object State Change Recognition": 0,
    "Ego Object Localization": 0,
    "Ego Object State Change Recognition": 0,
    "Action Recognition": 0,
    "Object Function": 0,
    "Information Function": 0,
    "Action Reasoning": 0,
    "Task Understanding": 0,
    
    
}

task2nonrecall_pred = {
    "Object Recognition": 0,
    "Attribute Perception": 0,
    "Text-Rich Understanding": 0,
    "Object Localization": 0,
    "Object State Change Recognition": 0,
    "Ego Object Localization": 0,
    "Ego Object State Change Recognition": 0,
    "Action Recognition": 0,
    "Object Function": 0,
    "Information Function": 0,
    "Action Reasoning": 0,
    "Task Understanding": 0,
    
}




total_fps_last = 0
total_fps = 0
total_response_number = 0
total_kv_cache_size = 0
for k,v in eval_result.items():
    for kk,vv in v.items():
        for ll in vv:
            if eval_model in ll.keys():
                for response in ll[eval_model]:
                    if response['role'] == 'fps':
                        total_fps_last+=response['content']
                    if response['role'].lower() == 'assistant':
                        if 'fps' in response:
                            total_fps+=response['fps']
                        total_response_number += 1
                        if 'kv_cache_size' in response:
                            total_kv_cache_size+=response['kv_cache_size']
                task2number[ll['Task Type'].strip()] += 1

print('total_qa: ', sum(task2number.values()))
print(f"Average FPS: {total_fps/total_response_number}")
print(f"Average KV Cache: {total_kv_cache_size/total_response_number}")
print("total_fps_last_mean: ", total_fps_last / sum(task2number.values()))
print(f"Average Response Number: {total_response_number/sum(task2number.values())}")