When I used SSD to train my own data set and read the train.txt file, the following problems occurred

1 
Traceback (most recent call last):
  File "D:/youyong/VS/deep-learning-for-image-processing-master/pytorch_object_detection/ssd/train_ssd300.py", line 181, in <module>
    main(args)
  File "D:/youyong/VS/deep-learning-for-image-processing-master/pytorch_object_detection/ssd/train_ssd300.py", line 114, in main
    val_data = get_coco_api_from_dataset(val_data_loader.dataset)
  File "D:\youyong\VS\deep-learning-for-image-processing-master\pytorch_object_detection\ssd\train_utils\coco_utils.py", line 66, in get_coco_api_from_dataset
    return convert_to_coco_api(dataset)
  File "D:\youyong\VS\deep-learning-for-image-processing-master\pytorch_object_detection\ssd\train_utils\coco_utils.py", line 17, in convert_to_coco_api
    targets = ds.coco_index(img_idx)
  File "D:\youyong\VS\deep-learning-for-image-processing-master\pytorch_object_detection\ssd\my_dataset.py", line 131, in coco_index
    xml_str = fid.read()
UnicodeDecodeError: 'gbk' codec can't decode byte 0xb9 in position 105: illegal multibyte sequence

How to solve it？

2

This error seems to be a decoding issue and might not be directly related to PyTorch.
Could you check the suggestions from this issue and see, if you could specify an encoding?