In version 0.x, in the load_data_list() function in basesegdataset.py and the encode_decode() function in encoder_decoder.py, the print information shows that it can read in sequence according to H48F020017_clip1_1_0_-1.png, H48F020017_clip1_2_0_-1 up to H48F020017_clip1_100_0_-1:

However, after upgrading to version 1.x, the load_data_list() function can still read in sequence, but the reading order of the encode_decode() function has changed to 100, 10, 11, 12…, and the order has become chaotic:

I want to try to add index conversion in the load_data_list() function, for example, convert 1 to 100 and 2 to 10, imagining whether it can be read correctly in the encode_decode() function. But I didn’t succeed, and I don’t think this is a good solution. I really want to know if the upgrade has changed the definitions of some functions, which has caused this problem. I hope someone can help me solve this problem. Thank you very much.

def load_data_list(self) -> List[dict]:
        """Load annotation from directory or annotation file.

        Returns:
            list[dict]: All data info of dataset.
        """
        data_list = []
        img_dir = self.data_prefix.get('img_path', None)
        ann_dir = self.data_prefix.get('seg_map_path', None)
        if not osp.isdir(self.ann_file) and self.ann_file:
            assert osp.isfile(self.ann_file), 
                f'Failed to load `ann_file` {self.ann_file}'
            lines = mmengine.list_from_file(
                self.ann_file, backend_args=self.backend_args)
            
            print(f'样本文件夹的数目: {len(lines)}')  # 打印 lines 数量
            
            for line in lines:
                img_name = line.strip()
                img_dir_path = osp.join(img_dir, img_name)  # 文件夹路径
                seg_dir_path = osp.join(ann_dir, img_name)  # 对应的标注文件夹路径
                
                img_files = glob.glob(osp.join(img_dir_path, '*' + self.img_suffix))  # 获取该文件夹中的所有影像文件
                seg_files = glob.glob(osp.join(seg_dir_path, '*' + self.seg_map_suffix))  # 获取该文件夹中的所有标注文件

                # 按文件名中的ID字段排序
                img_files = sort_by_natural_order(img_files)
                seg_files = sort_by_natural_order(seg_files)
               
                # print(f'Number of image files in {img_dir_path}: {len(img_files)}')  # 打印 img_files 数量
                # print(f'Number of segmentation files in {seg_dir_path}: {len(seg_files)}')  # 打印 seg_files 数量

                for img_file, seg_file in zip(img_files, seg_files):
                    #print(os.path.basename(img_file))
                    data_info = dict(
                        img_path=img_file,
                        seg_map_path=seg_file
                    ) 
                        
                    #todo：后面三个dict对应一个文件夹都是一样的，不知道有没有影响
                    data_info['label_map'] = self.label_map
                    data_info['reduce_zero_label'] = self.reduce_zero_label
                    data_info['seg_fields'] = []
                    data_list.append(data_info)
                    print(f'if:总共读取了 {len(data_list)} 个样本')
                    print('222222222222222222')  #--songyujian
                    print(f'Reading files from: {img_dir_path}')
                    print(f'Image files: {img_files}')
                    print(f'Segmentation files: {seg_files}')
        else:
            _suffix_len = len(self.img_suffix)
            for img in fileio.list_dir_or_file(
                    dir_path=img_dir,
                    list_dir=False,
                    suffix=self.img_suffix,
                    recursive=True,
                    backend_args=self.backend_args):
                data_info = dict(img_path=osp.join(img_dir, img))
                if ann_dir is not None:
                    seg_map = img[:-_suffix_len] + self.seg_map_suffix
                    data_info['seg_map_path'] = osp.join(ann_dir, seg_map)
                data_info['label_map'] = self.label_map
                data_info['reduce_zero_label'] = self.reduce_zero_label
                data_info['seg_fields'] = []
                data_list.append(data_info)
            
        data_list = sorted(data_list, key=lambda x: x['img_path'])
        return data_list

 def encode_decode(self, inputs: Tensor,
                      batch_img_metas: List[dict], overlap) -> Tensor:
        """Encode images with backbone and decode into a semantic segmentation
        map of the same size as input."""
        # 改
        print(batch_img_metas)
        print('!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!')
        x = self.extract_feat(inputs, overlap)
        # 改 predict 返回2个值，只接收前1个
        seg_logits = self.decode_head.predict(x, batch_img_metas,
                                               self.test_cfg)[0]
        #原来的：
        # seg_logits = self.decode_head.predict(x, batch_img_metas,
        #        

        return seg_logits