×
分享到微信

打开微信,使用扫一扫进入页面后,点击右上角菜单,

点击“发送给朋友”或“分享到朋友圈”完成分享

如何在docker下查看yolov5的推理效果 已完结 ZheJiangLab2022-06-27 16:05:59 回复 2 查看 技术答疑 使用求助
如何在docker下查看yolov5的推理效果
分享到:

您好!我按照Cambricon PyTorch Docker Image指引,启动docher

docker run -it --privileged --net=host  cair.cambricon.com/cambricon/cambricon_pytorch:ubuntu18.04_sdk_v1.7.0_pytorch_v0.15.0-2,然后执行在examples/offline/yolov5路径下yolov5的推理指令:

../build/yolov5/yolov5_offline_multicore ‐offlinemodel  yolov5.cambricon ‐dataset_path $COCO_PATH_PYTORCH ‐images ../../data/coco/val2017/file_list_for_release ‐labels label_map_coco.txt  ‐simple_compile 1 ‐input_format 0

报错如下:

CNRT: 4.10.1 a884a9a
load file:
2022-06-27 15:46:59.726429: [cnrtError] [156] [Card : 0] Open file failed! Error no: 2 - No such file or directory.
2022-06-27 15:46:59.726465: [cnrtError] [156] [Card : 0] Load model from file failed!
2022-06-27 15:46:59.726496: [cnrtError] [156] [Card : 0] cnrtExtractFunction: The pointer of model or pfunction is null, please check parameters.
Init runtime context for device0
2022-06-27 15:46:59.726535: [cnrtError] [156] [Card : 0] cnrtCreateRuntimeContext: Fail to create runtime context. Function is not initialized yet.
Failed to create runtime context
能给指点下,问题出在哪里吗?多谢!
是加载yolov5.cambricon模型失败吧,但是我在 /torch/examples/offline/执行了    ./ s/build_offline.sh。

版权所有 © 2022 寒武纪 Cambricon.com 备案/许可证号:京ICP备17003415号-1
关闭