lhz-1

> Traceback (most recent call last): File "D:/RunPrograms/edgeai-yolov5-pose/detect.py", line 219, in detect(opt=opt) File "D:/RunPrograms/edgeai-yolov5-pose/detect.py", line 111, in detect scale_coords(img.shape[2:], det[:, 6:], im0.shape, kpt_label=kpt_label, step=3) File "D:\RunPrograms\edgeai-yolov5-pose\utils\general.py", line 385, in scale_coords...

for det_index, (*xyxy, conf, cls) in enumerate(reversed(det[:,:6])): The value of 109 rows “reversed(det[:,:6])” makes the index different from the “det”，resulting in 119 rows“kpts”and“xyxy” not being the same data

https://blog.csdn.net/qq_37085158/article/details/132598829?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522169915164016800213054460%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fall.%2522%257D&request_id=169915164016800213054460&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~all~first_rank_ecpm_v1~rank_v31_ecpm-1-132598829-null-null.142^v96^pc_search_result_base1&utm_term=%20The%20GPU%20architecture%20in%20your%20current%20machine%20is%20Pascal%2C&spm=1018.2226.3001.4187 选择版本可以直接在[https://www.paddlepaddle.org.cn/whl/linux/mkl/avx/stable.html](https://www.paddlepaddle.org.cn/whl/linux/mkl/avx/stable.html) 解决办法 我是cuda11.6 cudnn 8.5.0 再笔记本上是2.5.0的rc1 可以使用 再另一台win10是2.5.0rc1 再centos上是2.4.2cuda版本和cudnn版本没变过 装完之后使用 ``` import paddle paddle.utils.run_check() ``` 只要不出现那个警告( GPU architecture in your current machine is Pascal, )应该可以了