多尺度测试

[sxlyiyiyi]

当我尝试使用多尺度测试时，出现以下错误

[06/05 21:01:13] reader WARNING: fail to map op [PadBatch_f39505] with error: list indices must be integers or slices, not str and stack: Traceback (most recent call last): File "E:\PaddleDetection\ppdet\data\reader.py", line 75, in call data = f(data) File "E:\PaddleDetection\ppdet\data\transform\batch_operators.py", line 62, in call max_shape = np.array([data['image'].shape for data in samples]).max( File "E:\PaddleDetection\ppdet\data\transform\batch_operators.py", line 62, in max_shape = np.array([data['image'].shape for data in samples]).max( TypeError: list indices must be integers or slices, not str

Exception in thread Thread-10: Traceback (most recent call last): File "C:\Users\sxl\anaconda3\envs\pd2\lib\threading.py", line 926, in _bootstrap_inner self.run() File "C:\Users\sxl\anaconda3\envs\pd2\lib\threading.py", line 870, in run self._target(*self._args, *self._kwargs) File "C:\Users\sxl\anaconda3\envs\pd2\lib\site-packages\paddle\fluid\dataloader\dataloader_iter.py", line 199, in _thread_loop six.reraise(sys.exc_info()) File "C:\Users\sxl\anaconda3\envs\pd2\lib\site-packages\six.py", line 719, in reraise raise value File "C:\Users\sxl\anaconda3\envs\pd2\lib\site-packages\paddle\fluid\dataloader\dataloader_iter.py", line 167, in _thread_loop batch = self._dataset_fetcher.fetch(indices) File "C:\Users\sxl\anaconda3\envs\pd2\lib\site-packages\paddle\fluid\dataloader\fetcher.py", line 66, in fetch data = self.collate_fn(data) File "E:\PaddleDetection\ppdet\data\reader.py", line 80, in call raise e File "E:\PaddleDetection\ppdet\data\reader.py", line 75, in call data = f(data) File "E:\PaddleDetection\ppdet\data\transform\batch_operators.py", line 62, in call max_shape = np.array([data['image'].shape for data in samples]).max( File "E:\PaddleDetection\ppdet\data\transform\batch_operators.py", line 62, in max_shape = np.array([data['image'].shape for data in samples]).max( TypeError: list indices must be integers or slices, not str

[yghstill]

@sxlyiyiyi 是release/2.0或者2.1中动态图模型使用MultiscaleTestResize时报错吗？动态图下多尺度测试支持还不完备，我们近期会高优支持下，可以暂时先使用单尺度预测。

[sxlyiyiyi]

@sxlyiyiyi 是release/2.0或者2.1中动态图模型使用MultiscaleTestResize时报错吗？动态图下多尺度测试支持还不完备，我们近期会高优支持下，可以暂时先使用单尺度预测。

是的，是在动态图模型下使用的