在jetsonnano上部属fastdeploy的python案例时，报错！！！！

温馨提示：根据社区不完全统计，按照模板提问，可以加快回复和解决问题的速度

环境

• 【FastDeploy版本】： 说明具体的版本，fastdeploy_gpu_python-0.0.0-cp36-cp36m-linux_aarch64.whl，在jetsonnano中编译的，下载最新的fastdeploy版本。
• 【编译命令】git clone https://github.com/PaddlePaddle/FastDeploy.git
  cd FastDeploy/python
  export BUILD_ON_JETSON=ON
  export ENABLE_VISION=ON
  python setup.py build
  python setup.py bdist_wheel
• 【系统平台】: Linux x64(Ubuntu 18.04)
• 【硬件】： 说明具体硬件型号，jetsonnano
• 【编译语言】：Python(3.6.9）

问题日志及出现问题的操作流程

• 附上详细的问题日志有助于快速定位分析

• 【模型跑不通】

• -在案例使用ppyooe模型在cpu、GPU时可以跑通，但使用trt时报错（python infer_ppyoloe.py --model_dir ppyoloe_crn_l_300e_coco --image 000000014439.jpg --device gpu --use_trt True），错误信息如下：

• [WARN][Paddle2ONNX] [multiclass_nms3: multiclass_nms3_0.tmp_1] Paramter nms_top_k:10000 is exceed limit in TensorRT BatchedNMS plugin, will force to 4096.
  [INFO] fastdeploy/runtime/backends/tensorrt/trt_backend.cc(572)::BuildTrtEngine Start to building TensorRT Engine...
  [INFO] fastdeploy/runtime/backends/tensorrt/trt_backend.cc(659)::BuildTrtEngine TensorRT Engine is built successfully.
  [INFO] fastdeploy/runtime/backends/tensorrt/trt_backend.cc(661)::BuildTrtEngine Serialize TensorRTEngine to local file ./tensorrt_cache/model.trt.
  [ERROR] fastdeploy/runtime/backends/tensorrt/trt_backend.cc(666)::BuildTrtEngine Failed to open ./tensorrt_cache/model.trt to write.
  [ERROR] fastdeploy/runtime/backends/tensorrt/trt_backend.cc(752)::CreateTrtEngineFromOnnx Failed to build tensorrt engine.
  [INFO] fastdeploy/runtime/runtime.cc(339)::CreateTrtBackend Runtime initialized with Backend::TRT in Device::GPU.
  [ERROR] fastdeploy/runtime/backends/tensorrt/trt_backend.cc(239)::log 2: [pluginV2DynamicExtRunner.cpp::execute::115] Error Code 2: Internal Error (Assertion status == kSTATUS_SUCCESS failed. )
  [ERROR] fastdeploy/runtime/backends/tensorrt/trt_backend.cc(348)::InferFailed to Infer with TensorRT.
  [ERROR] fastdeploy/vision/detection/ppdet/base.cc(73)::BatchPredict Failed to inference by runtime

• 第二个问题是运行案例中的ppyolo模型报错（python infer_ppyolo.py），报错信息如下：

• dlinano@jetson-nano:~/FastDeploy/examples/vision/detection/paddledetection/python$ python infer_ppyolo.py
  100%|█████████████████████████| 171084/171084 [00:20<00:00, 8417.70KB/s]
  Successfully download model at path: /home/dlinano/.fastdeploy/models/ppyolo_r50vd_dcn_1x_coco
  [INFO] fastdeploy/vision/common/processors/transform.cc(45)::FuseNormalizeCast Normalize and Cast are fused to Normalize in preprocessing pipeline.
  [INFO] fastdeploy/vision/common/processors/transform.cc(93)::FuseNormalizeHWC2CHW Normalize and HWC2CHW are fused to NormalizeAndPermute in preprocessing pipeline.
  [INFO] fastdeploy/vision/common/processors/transform.cc(159)::FuseNormalizeColorConvert BGR2RGB and NormalizeAndPermute are fused to NormalizeAndPermute with swap_rb=1
  [ERROR] fastdeploy/fastdeploy_model.cc(239)::CreateCpuBackend Found no valid backend for model: PaddleDetection/PP-YOLO
  [ERROR] fastdeploy/vision/detection/ppdet/base.cc(32)::Initialize Failed to initialize fastdeploy backend.
  Traceback (most recent call last):
  File "infer_ppyolo.py", line 62, in
  model_file, params_file, config_file, runtime_option=runtime_option)
  File "/usr/local/lib/python3.6/dist-packages/fastdeploy/vision/detection/ppdet/init.py", line 188, in init
  assert self.initialized, "PPYOLO model initialize failed."
  AssertionError: PPYOLO model initialize failed.

Start with examples/vision/detection/paddledetection/python/infer_ppyolo.py and reproduce both reported paths: CPU/GPU versus TensorRT, and the default PP-YOLO run. Read the logged TensorRT failure in runtime/backends/tensorrt/trt_backend.cc and the backend initialization failure, then verify that both inference paths initialize and complete successfully on the stated Jetson Nano setup.