描述:
将本地训练好的yolov5的模型(分类、分割、检测)多机分离部署,传给客户环境远程调试与测试
提示:
tensorrt移植时,多机的显卡算力需要和训练模型机器的显卡算力一致
问题1:
Load tensorrt engine文件出错\\engine plan file is not compatible with this version of TensorRT 8.4.15 got 8.4
多机上Tensorrt的设置需要和生成模型时单机的tensorrt设置一致
问题2:
Could not locate zlibwapi.dll. Please make sure it is in your library path!
cudnn的bin、lib、include放到cuda的目录下
问题3:
Code 2: Internal Error (Assertion validateCaskKLibSize(buffer.size) failed