feat(actpred): 训练时安全图像增强（ColorJitter），支持任意 frame_len

训练集启用 --augment 后对每帧施加亮度/对比度/饱和度/色调变换，
同一样本内所有帧共享参数（模拟连续帧光照不变），验证集不增强。
仅颜色空间变换，无空间变换，动作标签不受影响。
删除 t1.mt（@TEST 块无对应插件，录制时报错）。

actpred/config.py

@@ -40,6 +40,7 @@ class TrainConfig:
     num_workers: int = 2
     seed: int = 42
     max_grad_norm: float = 1.0
+    augment: bool = False
 
 
 @dataclass
@@ -87,6 +88,8 @@ def build_argparser(description: str = "") -> argparse.ArgumentParser:
     parser.add_argument("--num-workers", type=int, default=None)
     parser.add_argument("--seed", type=int, default=None)
     parser.add_argument("--max-grad-norm", type=float, default=None)
+    parser.add_argument("--augment", action=argparse.BooleanOptionalAction, default=None,
+                        help="训练时启用安全图像增强（ColorJitter：亮度/对比度/饱和度/色调）")
     parser.add_argument("--init-from", type=str, default=None,
                         help="预训练 checkpoint 路径，作为微调起点（strict=False 加载兼容权重）")
     parser.add_argument("--resume", type=str, default=None,
@@ -232,6 +235,7 @@ def apply_overrides(cfg: Config, args: argparse.Namespace) -> Config:
         "num_workers": ("train", "num_workers"),
         "seed": ("train", "seed"),
         "max_grad_norm": ("train", "max_grad_norm"),
+        "augment": ("train", "augment"),
     }
     args_dict = vars(args)
     for arg_name, (section, field) in mapping.items():

actpred/dataset.py

@@ -1,4 +1,5 @@
 import os
+import random
 import re
 import numpy as np
 from pathlib import Path
@@ -7,6 +8,7 @@ from PIL import Image
 import torch
 from torch.utils.data import Dataset
 from torchvision import transforms
+from torchvision.transforms import functional as TF
 
 
 def _natural_key(p):
@@ -19,12 +21,13 @@ class ActionDataset(Dataset):
                  image_width: int = 640, image_height: int = 480,
                  normalize_actions: bool = False,
                  action_stats_path: str = None, episode_indices: list = None,
-                 frame_len: int = 3):
+                 frame_len: int = 3, augment: bool = False):
         self.history_len = history_len
         self.action_dim = action_dim
         self.image_width = image_width
         self.image_height = image_height
         self.frame_len = frame_len
+        self.augment = augment
 
         self.transform = transforms.Compose([
             transforms.Resize((image_height, image_width)),
@@ -92,8 +95,18 @@ class ActionDataset(Dataset):
     def __getitem__(self, idx):
         image_paths, action_history, target = self.samples[idx]
         frames = []
+        if self.augment:
+            bf = random.uniform(0.8, 1.2)
+            cf = random.uniform(0.8, 1.2)
+            sf = random.uniform(0.8, 1.2)
+            hf = random.uniform(-0.05, 0.05)
         for p in image_paths:
             image = Image.open(p).convert("RGB")
+            if self.augment:
+                image = TF.adjust_brightness(image, bf)
+                image = TF.adjust_contrast(image, cf)
+                image = TF.adjust_saturation(image, sf)
+                image = TF.adjust_hue(image, hf)
             frames.append(self.transform(image))
         images = torch.stack(frames)  # (F, 3, S, S)
         action_history = torch.tensor(action_history, dtype=torch.float32)

actpred/train.py

@@ -84,7 +84,8 @@ def main():
     train_ds = ActionDataset(
         cfg.data.data_dir, cfg.data.history_len, cfg.data.action_dim,
         cfg.data.image_width, cfg.data.image_height, cfg.data.normalize_actions,
-        cfg.data.action_stats_path, train_indices, cfg.data.frame_len)
+        cfg.data.action_stats_path, train_indices, cfg.data.frame_len,
+        augment=cfg.train.augment)
     if cfg.data.normalize_actions and not cfg.data.action_stats_path:
         # 归一化统计量必须只用训练集计算，再供验证集复用，否则 val loss 不可信
         stats_path = os.path.join(cfg.train.checkpoint_dir, "action_stats.npy")