faster_rcnn训练自己的数据集训练一个Epoch后报错

Traceback (most recent call last):
  File "train_res50_fpn.py", line 231, in <module>
    main(args)
  File "train_res50_fpn.py", line 153, in main
    coco_info = utils.evaluate(model, val_data_set_loader, device=device)
  File "D:\App\miniconda3\envs\faster_rcnn\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "D:\GitClone\faster_rcnn\train_utils\train_eval_utils.py", line 76, in evaluate
    coco = get_coco_api_from_dataset(data_loader.dataset)
  File "D:\GitClone\faster_rcnn\train_utils\coco_utils.py", line 54, in get_coco_api_from_dataset
    return convert_to_coco_api(dataset)
  File "D:\GitClone\faster_rcnn\train_utils\coco_utils.py", line 15, in convert_to_coco_api
    hw, targets = ds.coco_index(img_idx)
AttributeError: 'VOCDataSet' object has no attribute 'coco_index'