使用trtexec命令行转换得到的trt推理报错

[Nolan0206]

首先非常感谢你的分享。
我在windows11下使用pip 安装nvidia-tensorrt会报错，所以在docker环境中使用trtexec命令进行转换，在用trt.py推理时遇到如下报错：
Namespace(end2end=False, engine='../yolov7/weights/weights/test_no_sim.trt', image='../yolov7/inference/images/boat.jpg', output='test2.jpg', video=None)
[09/01/2022-03:26:21] [TRT] [E] 1: [stdArchiveReader.cpp::StdArchiveReader::35] Error Code 1: Serialization (Serialization assertion safeVersionRead == safeSerializationVersion failed.Version tag does not match. Note: Current Version: 0, Serialized Engine Version: 43)
[09/01/2022-03:26:21] [TRT] [E] 4: [runtime.cpp::deserializeCudaEngine::50] Error Code 4: Internal Error (Engine deserialization failed.)
Traceback (most recent call last):
File "trt.py", line 27, in
pred = Predictor(engine_path=args.engine)
File "trt.py", line 11, in init
super(Predictor, self).init(engine_path)
File "/mnt/tensorrt-python/utils/utils.py", line 31, in init
self.context = engine.create_execution_context()
AttributeError: 'NoneType' object has no attribute 'create_execution_context'
使用的是自己训练的yolov7，转onnx命令是：python export.py --weights weights/weights/best_rep.pt
转trt命令是/usr/src/tensorrt/bin/trtexec --onnx=best_rep.onnx --saveEngine=test_no_sim.trt --fp16
推理命令是python trt.py -e ../yolov7/weights/weights/test_no_sim.trt -i ../yolov7/inference/images/boat.jpg -o test2.jpg
docker环境是参照yolov7和你的pip安装搭建的，可以跑end2end，trt的版本不存在问题。

[Nolan0206]

@Linaom1214

[Linaom1214]

docker 里trt版本和本地
版本不一致

 Error Code 1: Serialization (Serialization assertion safeVersionRead == safeSerializationVersion failed.Version tag does not match. Note: Current Version: 0, Serialized Engine Version: 43)

[rance1108]

[E] Error[1]: [stdArchiveReader.cpp::StdArchiveReader::34] Error Code 1: Serialization (Serialization assertion safeVersionRead == safeSerializationVersion failed.Version tag does not match. Note: Current Version: 43, Serialized Engine Version: 0)
I got the same error, which version of docker image or tensorrt should I use?

[Linaom1214]

Current Version: 43, Serialized Engine Versi

the export engine env should same as the infer env