layout fastdeploy_serving方式部署 Failed to process the request(s) for model 'postprocess_0', message: TypeError: run(): incompatible function arguments.

[lsx66]

版面分析算法, fastdeploy_serving方式部署,在StructureV2LayoutPostprocessor后处理时报错
Failed to process the request(s) for model 'postprocess_0', message: TypeError: run(): incompatible function arguments. The following argument types are supported:
1. (self: fastdeploy.libs.fastdeploy_main.vision.ocr.StructureV2LayoutPostprocessor, arg0: List[fastdeploy.libs.fastdeploy_main.FDTensor], arg1: List[List[int[4]]]) -> List[fastdeploy.libs.fastdeploy_main.vision.DetectionResult]
Invoked with: <fastdeploy.libs.fastdeploy_main.vision.ocr.StructureV2LayoutPostprocessor object at 0x7f788b5eb4f0>, array([[[ 3.0430562 , 2.9620893 , -0.92005384, ..., -1.1082802 ,
-1.694324 , -1.3905903 ],
[-0.8745545 , 3.1247966 , 3.0257077 , ..., -1.0160857 ,
-1.3037924 , -0.8528714 ],
[-1.1111182 , -1.1563158 , 3.6807263 , ..., -0.8917918 ,
-1.5667745 , -1.2615712 ],
...,
[ 2.574394 , 2.0140955 , -0.02018397, ..., -1.5280299 ,
-1.9861596 , -1.8610322 ],
[ 0.99233127, 1.3739115 , 0.8655137 , ..., -1.4873394 ,
-1.9895219 , -1.8228513 ],
[ 0.2792613 , 0.42100033, 1.0559945 , ..., -1.313761 ,
-1.8969777 , -1.7097901 ]]], dtype=float32), [array([640, 404, 608, 800], dtype=object)]

At:
/usr/local/lib/python3.8/dist-packages/fastdeploy/vision/ocr/ppocr/init.py(778): run
/ocr_serving/vision/ocr/PP-OCR/serving/fastdeploy_serving/models-layout/postprocess/1/model.py(109): execute

版面分析后处理model.py报错行
results = self.postprocessor_.run([infer_outputs], [im_infos])
参数打印出
infer_outputs shape is (1, 130, 32)
im_infos: [640 404 608 800]

ppocr/init.py 中StructureV2LayoutPostprocessor类
class StructureV2LayoutPostprocessor:
def init(self):
"""Create a postprocessor for StructureV2Layout Model
"""
self._postprocessor = C.vision.ocr.StructureV2LayoutPostprocessor()

def run(self, runtime_results, ims_info):
    """Postprocess the runtime results for StructureV2Layout Model
    :param: runtime_results: (list of FDTensor or list of pyArray)The output FDTensor results from runtime
    :return: list of Result(If the runtime_results is predict by batched samples, the length of this list equals to the batch size)
    """
    return self._postprocessor.run(runtime_results, ims_info)

[Jiang-Jia-Jun]

具体报错日志截图发下

[lsx66]

[Jiang-Jia-Jun]

看这个提示应该是 这个函数第一个参数要求是FDTensor，但实际上传入了一个np.ndarray