Add files using upload-large-folder tool

extract_h5.py

@@ -1,30 +1,61 @@
 #!/usr/bin/env python3
 """
 解析 H5 文件并导出为 Hugging Face Dataset Viewer 兼容格式
-按帧展开，每行一帧数据
+
+用法:
+  python extract_h5.py          # 解析所有 H5 文件
+  python extract_h5.py --check  # 仅检查 H5 文件结构（不解析）
 """
 
+import argparse
 import h5py
 import numpy as np
 from pathlib import Path
 from PIL import Image
 from tqdm import tqdm
 import json
-import matplotlib
-matplotlib.use('Agg')  # 无 GUI 后端
-import matplotlib.pyplot as plt
+from collections import defaultdict
 
 
-def save_heatmap(data, output_path, cmap='viridis'):
-    """将矩阵保存为热力图 PNG"""
-    fig, ax = plt.subplots(figsize=(6, 4))
-    im = ax.imshow(data, cmap=cmap, aspect='auto')
-    plt.colorbar(im, ax=ax)
-    ax.set_xlabel('Column')
-    ax.set_ylabel('Row')
-    plt.tight_layout()
-    plt.savefig(output_path, dpi=100)
-    plt.close(fig)
+def check_h5_structure(base_dir):
+    """检查所有 H5 文件的字段结构"""
+    folder_keys = defaultdict(lambda: defaultdict(set))
+    
+    h5_folders = [d for d in base_dir.iterdir() if d.is_dir() and d.name.endswith('_h5')]
+    
+    for h5_folder in sorted(h5_folders):
+        h5_files = list(h5_folder.rglob('*.h5'))
+        print(f"\n{'='*60}")
+        print(f"文件夹: {h5_folder.name} ({len(h5_files)} 个文件)")
+        print('='*60)
+        
+        # 检查前 3 个文件作为示例
+        for h5_path in h5_files[:3]:
+            print(f"\n  {h5_path.name}:")
+            with h5py.File(h5_path, 'r') as f:
+                for key in sorted(f.keys()):
+                    arr = f[key]
+                    print(f"    - {key}: shape={arr.shape}, dtype={arr.dtype}")
+                    folder_keys[h5_folder.name][key].add(str(arr.shape))
+        
+        # 汇总该文件夹的所有键
+        print(f"\n  汇总 (检查全部 {len(h5_files)} 个文件):")
+        for h5_path in h5_files:
+            with h5py.File(h5_path, 'r') as f:
+                for key in f.keys():
+                    folder_keys[h5_folder.name][key].add(str(f[key].shape))
+        
+        for key, shapes in sorted(folder_keys[h5_folder.name].items()):
+            print(f"    - {key}: shapes={list(shapes)}")
+    
+    # 总汇总
+    print(f"\n{'='*60}")
+    print("总汇总 - 所有文件夹的字段:")
+    print('='*60)
+    for folder, keys in sorted(folder_keys.items()):
+        print(f"\n{folder}:")
+        for key, shapes in sorted(keys.items()):
+            print(f"  - {key}: {list(shapes)}")
 
 
 def extract_pose_data(h5_path, output_dir, episode_id, subset_path=""):
@@ -42,7 +73,7 @@ def extract_pose_data(h5_path, output_dir, episode_id, subset_path=""):
         
         # 预处理所有数据
         data_cache = {}
-        image_paths = {}  # key -> list of paths per frame
+        image_paths = {}
         
         for key in keys:
             arr = f[key][:]
@@ -63,7 +94,7 @@ def extract_pose_data(h5_path, output_dir, episode_id, subset_path=""):
                     
                 elif len(arr.shape) == 5:  # 多视角 (N, V, H, W, C)
                     num_views = arr.shape[1]
-                    paths = []  # 每帧的多视角路径
+                    paths = []
                     for frame_idx in range(arr.shape[0]):
                         frame_paths = []
                         for view_idx in range(num_views):
@@ -86,18 +117,18 @@ def extract_pose_data(h5_path, output_dir, episode_id, subset_path=""):
             if subset_path:
                 record["subset"] = subset_path
             
-            # 添加该帧的图像路径 (使用 file_name 作为主图像列，符合 HF 格式)
+            # 添加该帧的图像路径 (file_name 为 HF 格式必需)
             for key, paths in image_paths.items():
                 if isinstance(paths[0], list):  # 多视角
                     for v_idx, p in enumerate(paths[frame_idx]):
                         if v_idx == 0:
-                            record["file_name"] = p  # 第一个视角作为主图像
+                            record["file_name"] = p
                         record[f"image_v{v_idx}"] = p
                 else:
-                    record["file_name"] = paths[frame_idx]  # 主图像
+                    record["file_name"] = paths[frame_idx]
                     record["image"] = paths[frame_idx]
             
-            # 添加静态数据（如背景图）
+            # 添加静态数据
             for key, val in data_cache.items():
                 if key.endswith("_image") or key.endswith("_num_views"):
                     record[key] = val
@@ -110,11 +141,8 @@ def extract_pose_data(h5_path, output_dir, episode_id, subset_path=""):
             if 'translations' in data_cache:
                 record["translation"] = data_cache['translations'][frame_idx]
             if 'tactile' in data_cache:
-                # tactile 数据直接放 JSON (tac02: 20x66)
                 record["tactile"] = data_cache['tactile'][frame_idx]
-            
             if 'xela' in data_cache:
-                # xela 数据直接放 JSON (xela: 10x72)
                 record["xela"] = data_cache['xela'][frame_idx]
             
             record["num_frames"] = num_frames
@@ -137,22 +165,19 @@ def extract_tacniq_gsmini(h5_path, output_dir, episode_id, subset_path=""):
         gsmini = f['gsmini'][:]
         tacniq = f['tacniq'][:].tolist()
     
-    # 保存背景图
     Image.fromarray(bg).save(episode_dir / "bg.png")
     bg_path = f"{rel_prefix}/bg.png"
     
     num_frames = len(gsmini)
     
-    # 按帧展开
     for frame_idx in range(num_frames):
-        # 保存该帧图像
         filename = f"gsmini_{frame_idx:04d}.png"
         Image.fromarray(gsmini[frame_idx]).save(episode_dir / filename)
         
         record = {
             "episode_id": episode_id,
             "frame_idx": frame_idx,
-            "file_name": f"{rel_prefix}/{filename}",  # HF 格式必需
+            "file_name": f"{rel_prefix}/{filename}",
             "image": f"{rel_prefix}/{filename}",
             "bg_image": bg_path,
             "tacniq": tacniq[frame_idx] if frame_idx < len(tacniq) else None,
@@ -181,22 +206,19 @@ def extract_xela_9dtact(h5_path, output_dir, episode_id, subset_path=""):
         dtact = f['9dtact'][:]
         xela = f['xela'][:].tolist()
     
-    # 保存背景图
     Image.fromarray(bg).save(episode_dir / "bg.png")
     bg_path = f"{rel_prefix}/bg.png"
     
     num_frames = len(dtact)
     
-    # 按帧展开
     for frame_idx in range(num_frames):
-        # 保存该帧图像
         filename = f"9dtact_{frame_idx:04d}.png"
         Image.fromarray(dtact[frame_idx]).save(episode_dir / filename)
         
         record = {
             "episode_id": episode_id,
             "frame_idx": frame_idx,
-            "file_name": f"{rel_prefix}/{filename}",  # HF 格式必需
+            "file_name": f"{rel_prefix}/{filename}",
             "image": f"{rel_prefix}/{filename}",
             "bg_image": bg_path,
             "xela": xela[frame_idx] if frame_idx < len(xela) else None,
@@ -223,7 +245,6 @@ def extract_force_data(h5_path, output_dir, episode_id, subset_path=""):
     with h5py.File(h5_path, 'r') as f:
         keys = list(f.keys())
         
-        # 找到帧数（通过时间戳或图像序列）
         num_frames = 0
         data_cache = {}
         image_paths = {}
@@ -232,11 +253,11 @@ def extract_force_data(h5_path, output_dir, episode_id, subset_path=""):
             arr = f[key][:]
             
             if arr.dtype == np.uint8:
-                if len(arr.shape) == 3:  # 背景图
+                if len(arr.shape) == 3:
                     filename = f"{key}.png"
                     Image.fromarray(arr).save(episode_dir / filename)
                     data_cache[f"{key}_image"] = f"{rel_prefix}/{filename}"
-                elif len(arr.shape) == 4:  # 图像序列
+                elif len(arr.shape) == 4:
                     num_frames = max(num_frames, len(arr))
                     paths = []
                     for i, img in enumerate(arr):
@@ -249,7 +270,6 @@ def extract_force_data(h5_path, output_dir, episode_id, subset_path=""):
                 if len(arr.shape) >= 1:
                     num_frames = max(num_frames, len(arr))
         
-        # 按帧展开
         for frame_idx in range(num_frames):
             record = {
                 "episode_id": episode_id,
@@ -260,13 +280,11 @@ def extract_force_data(h5_path, output_dir, episode_id, subset_path=""):
             if subset_path:
                 record["subset"] = subset_path
             
-            # 添加图像路径
             for key, paths in image_paths.items():
                 if frame_idx < len(paths):
-                    record["file_name"] = paths[frame_idx]  # HF 格式必需
+                    record["file_name"] = paths[frame_idx]
                     record["image"] = paths[frame_idx]
             
-            # 添加静态数据
             for key, val in data_cache.items():
                 if key.endswith("_image"):
                     record[key] = val
@@ -278,25 +296,20 @@ def extract_force_data(h5_path, output_dir, episode_id, subset_path=""):
     return records
 
 
-def main():
-    base_dir = Path(__file__).parent
-    
-    # 找到所有 *_h5 文件夹
+def extract_all(base_dir):
+    """解析所有 H5 文件"""
     h5_folders = [d for d in base_dir.iterdir() if d.is_dir() and d.name.endswith('_h5')]
     
     for h5_folder in h5_folders:
-        # 创建对应的输出文件夹（去掉 _h5 后缀）
         output_folder = base_dir / h5_folder.name.replace('_h5', '')
         output_folder.mkdir(exist_ok=True)
         
-        # 获取所有 h5 文件（包括子目录）
         h5_files = list(h5_folder.rglob('*.h5'))
         print(f"\n处理 {h5_folder.name}: {len(h5_files)} 个文件 -> {output_folder.name}/")
         
         all_records = []
         
         for h5_path in tqdm(h5_files, desc=h5_folder.name):
-            # 计算相对路径，保持子目录结构
             relative = h5_path.relative_to(h5_folder)
             sub_output_dir = output_folder / relative.parent
             sub_output_dir.mkdir(parents=True, exist_ok=True)
@@ -318,7 +331,6 @@ def main():
                 
                 all_records.extend(records)
                 
-                # 保存该 episode 的 JSON 文件
                 episode_dir = sub_output_dir / episode_id
                 json_path = episode_dir / "metadata.json"
                 with open(json_path, 'w') as f:
@@ -327,7 +339,6 @@ def main():
             except Exception as e:
                 print(f"\nError: {h5_path}: {e}")
         
-        # 保存 metadata.jsonl（Hugging Face Dataset Viewer 格式）
         jsonl_path = output_folder / "metadata.jsonl"
         with open(jsonl_path, 'w') as f:
             for record in all_records:
@@ -338,5 +349,18 @@ def main():
     print("\n解析完成！")
 
 
+def main():
+    parser = argparse.ArgumentParser(description="解析 H5 文件为 HuggingFace Dataset 格式")
+    parser.add_argument('--check', action='store_true', help='仅检查 H5 文件结构（不解析）')
+    args = parser.parse_args()
+    
+    base_dir = Path(__file__).parent
+    
+    if args.check:
+        check_h5_structure(base_dir)
+    else:
+        extract_all(base_dir)
+
+
 if __name__ == "__main__":
     main()