yolov5.cpp:245: void APIToModel(unsigned int, nvinfer1::IHostMemory**, bool&, float&, float&, std::__cxx11::string&): Assertion `engine != nullptr' failed. 已放弃

[goodworth1]

Env

• GPU, Xavier NX, .
• OS, e.g. Ubuntu18.04,Jetpack 4.6.1
• Cuda version 10.2.300
• TensorRT version 8.2.1.8

About this repo

• which model? yolov5, retinaface?
• yolov5-5.0

Your problem

我用的是官方下载的yolov5-5.0，在生成引擎文件时出现问题，yolov5s.pt是从官方下载的，按照https://github.com/wang-xinyu/tensorrtx/tree/yolov5-v5.0/yolov5操作生成引擎文件在sudo ./yolov5 -s yolov5s.wts yolov5s.engine s遇到问题

[wang-xinyu]

Strange error: Unknown embeded device detected.

Can you try other device/platform?

[XiantaoHu]

请问你是转换为int8吗

[StrongandSurvive]

请问您解决这个问题了吗，我在Xavier NX 16G 版本（Ubuntu18.04,Jetpack 4.6.2,Cuda version 10.2.300,TensorRT version 8.2.1.8）也遇到了同样的问题

[Taikonaut-llk]

also Xavier NX 16G .I also encountered this problem, it is very difficult, how can I solve it?

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[arielsolomon]

Same problem with Jetson orin model quantization: void APIToModel(unsigned int, nvinfer1::IHostMemory**, bool&, float&, float&, std::string&): Assertion `engine != nullptr' failed

[wang-xinyu]

@arielsolomon Did you try FP32 or FP16?

[arielsolomon]

No, will do I did try however with both your tutorial (meaning suggested yv5 version and yv5s model and data) and with my custom model and data. Both did not work It is an opportunity to thank your hard work!

On Mon, Nov 28, 2022 at 4:35 AM Wang Xinyu @.***> wrote:

@arielsolomon https://github.com/arielsolomon Did you try FP32 or FP16?

— Reply to this email directly, view it on GitHub https://github.com/wang-xinyu/tensorrtx/issues/998#issuecomment-1328450003, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALCFNJALDZTEEG5QO6QNXSLWKQK6DANCNFSM5VQ4NMHA . You are receiving this because you were mentioned.Message ID: @.***>

-- Dr. Ariel Solomon Mobile: 972-545-667301

[arielsolomon]

Sorry, wrong answer before, FP16 is the default so ofcourse I tried it and it was fine

On Mon, Nov 28, 2022 at 4:35 AM Wang Xinyu @.***> wrote:

@arielsolomon https://github.com/arielsolomon Did you try FP32 or FP16?

— Reply to this email directly, view it on GitHub https://github.com/wang-xinyu/tensorrtx/issues/998#issuecomment-1328450003, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALCFNJALDZTEEG5QO6QNXSLWKQK6DANCNFSM5VQ4NMHA . You are receiving this because you were mentioned.Message ID: @.***>

-- Dr. Ariel Solomon Mobile: 972-545-667301

[wang-xinyu]

Okay, INT8 requires calibration, did you follow this https://github.com/wang-xinyu/tensorrtx/tree/master/yolov5#int8-quantization ?

[arielsolomon]

Step by step! Again both same as your tutorial (yolov5s model) and with my custom model and custom data

On Tue, Nov 29, 2022 at 4:26 AM Wang Xinyu @.***> wrote:

Okay, INT8 requires calibration, did you follow this https://github.com/wang-xinyu/tensorrtx/tree/master/yolov5#int8-quantization ?

— Reply to this email directly, view it on GitHub https://github.com/wang-xinyu/tensorrtx/issues/998#issuecomment-1329992706, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALCFNJGV473RMAAJLD6HUV3WKVSVXANCNFSM5VQ4NMHA . You are receiving this because you were mentioned.Message ID: @.***>

-- Dr. Ariel Solomon Mobile: 972-545-667301

[wang-xinyu]

@arielsolomon Can you please open a new thread and paste all the details and logs?

[arielsolomon]

will do, thanks

בתאריך יום ד׳, 30 בנוב׳ 2022, 6:27, מאת Wang Xinyu ‏< @.***>:

@arielsolomon https://github.com/arielsolomon Can you please open a new thread and paste all the details and logs?

— Reply to this email directly, view it on GitHub https://github.com/wang-xinyu/tensorrtx/issues/998#issuecomment-1331629412, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALCFNJG5TQKINH3GUXXB5H3WK3JTBANCNFSM5VQ4NMHA . You are receiving this because you were mentioned.Message ID: @.***>