yolov5-tensorrt
yolov5-tensorrt copied to clipboard
|yolov5|error|[Detector] loadEngine() failure: could not deserialize engine
Hi
I have below env
GPU: Xavier NX OS: Ubuntu18.04 Cuda version: 10.2.300 TensorRT version: 8.0.1.6
I have downloaded and converted the yolov5s.pt
into yolov5s.wts
and yolov5s.engine
file using https://github.com/wang-xinyu/tensorrtx
I have been able to do inferencing on images using engine file. I have downloaded and installed this repo and library. I am running the process_live example by using below command:
python3 process_live.py --engine yolov5s.engine
but getting below error
|yolov5|info|[Detector] Using OpenCV-CPU pre-processor
|yolov5|info|[TensorRT] [MemUsageChange] Init CUDA: CPU +353, GPU +0, now: CPU 377, GPU 5657 (MiB)
|yolov5|info|[Detector] Loading TensorRT engine from '/home/john/Documents/tensorrtx/yolov5/build/yolov5s.engine'
|yolov5|info|[Detector] Deserializing inference engine. This may take a while...
|yolov5|info|[TensorRT] Loaded engine size: 18 MB
|yolov5|info|[TensorRT] [MemUsageSnapshot] deserializeCudaEngine begin: CPU 395 MiB, GPU 5675 MiB
|yolov5|error|[TensorRT] 3: getPluginCreator could not find plugin: YoloLayer_TRT version: 1
|yolov5|error|[TensorRT] 1: [pluginV2Runner.cpp::load::292] Error Code 1: Serialization (Serialization assertion creator failed.Cannot deserialize plugin since corresponding IPluginCreator not found in Plugin Registry)
|yolov5|error|[TensorRT] 4: [runtime.cpp::deserializeCudaEngine::76] Error Code 4: Internal Error (Engine deserialization failed.)
|yolov5|error|[Detector] loadEngine() failure: could not deserialize engine
loadEngine() failed: tensorrt error
Please suggest how to resolve this error. Thanks