Jason Yuan
Jason Yuan
You need to upgrade your graphics driver，my Driver Version: 418.39
> What is your graphics card model? p100
no,i have not , it also has some problems,When I execute the line 139(opencv=4.0.0)
thank you very much!
@jacquesmats Hello ,I have the same problems , the line 139 opencv=4.0.0 ,There is a conflict between opencv and gdal
检查路径。路径错了，生成为空
> I downgraded the version of paddlepaddle to 2.5.2 and the issue was solved > > 降低 paddlepaddle的版本到2.5.2，问题解决 1.实体机亲测可以解决，2.虚拟机在跑demo文件时，如果使用for循环ocr.ocr识别完成后貌似不会释放内存，连续识别图像，前几张可以，后几张报错“Process finished with exit code 137 (interrupted by signal 9:SIGKILL)”。 内存不足杀进程了
> 弱的多问一句，GPU版本的怎么安装？T4卡的服务器，直接按安装流程，上去就可以torch.cuda.is_available() 就是True，结果我本地win10，有cuda12.2的，同样安装好后，就是False，不知道怎么办了。 你需要nvidia docker
[1844301011125D0F00-depthai_crash_dump(1).json](https://github.com/luxonis/depthai-core/files/15451357/1844301011125D0F00-depthai_crash_dump.1.json) Thank you so much！this is the bug report file。