Closed Ruid6 closed 6 months ago
Did you use trtexec to export your model?
i use tools/deploy.py.
change config file batch input.
backend_config = dict( common_config=dict(max_workspace_size=1 << 30), model_inputs=[ dict( input_shapes=dict( input=dict( min_shape=[1, 3, 320, 320], opt_shape=[2, 3, 320, 320], max_shape=[4, 3, 320, 320]))) ])
it work in python. but here it overflow in c++.
IExecutionContext* createExecutionContext() noexcept { return mImpl->createExecutionContext(); }
print 3: getPluginCreator could not find plugin: TRTBatchedNMS version: 1 1: [pluginV2Runner.cpp::nvinfer1::rt::load::303] Error Code 1: Serialization (Serialization assertion creator failed.Cannot deserialize plugin since corresponding IPluginCreator not found in Plugin Registry)
Try to use convert_rtmdet.py to export onnx and use trtexec to get engine file.
Do not use mmdeploy to export the model as there are operators that are not supported
use convert_rtmdet.py to export both det and pose ? or only det?
only rtmdet
i use convert_rtmdet.py .i have two questions. 1.it work on my cv::VideoCapture cap(1);,but not real time. 2.i want batch inference rtmpose,what do i need?
i use convert_rtmdet.py .i have two questions. 1.it work on my cv::VideoCapture cap(1);,but not real time. 2.i want batch inference rtmpose,what do i need?
Is the problem resolved?
my card is 1050, I haven't tried it yet because I'm not too familiar with tensorrt, I use pose_trackers.cpp in mmdepoly.
my card is 1050, I haven't tried it yet because I'm not too familiar with tensorrt, I use pose_trackers.cpp in mmdepoly.
ok, If you have any questions, please raise them in the issue
I had the same problem. Can you please tell me how you finally solved it?
I have an error using your code, how should I fix it? my environment : TensorRT 8.6.0 Windows 10 VS 2022