PP-HumanSeg: IndexError: too many indices for array: array is 3-dimensional, but 4 were indexed

[andy315]

问题确认 Search before asking

• [X] 我已经查询历史issue(包括open与closed)，没有发现相似的bug。I have searched the open and closed issues and found no similar bug report.

Bug描述 Describe the Bug

训练微调之后得到的model（human_pp_humansegv2_lite）, 运行推理的时候报错： Traceback (most recent call last): File "C:\Python38\lib\tkinter_init_.py", line 1892, in call return self.func(*args) File "C:\Python38\lib\tkinter_init_.py", line 814, in callit func(*args) File ".\src\test.py", line 125, in main out = predictor.run(bgr_image, bg) File "D:\Workplace\PaddleSeg\contrib\PP-HumanSeg\src\infer.py", line 122, in run return self.postprocess(output, img, data, bg) File "D:\Workplace\PaddleSeg\contrib\PP-HumanSeg\src\infer.py", line 126, in postprocess score_map = pred_img[0, 1, :, :] IndexError: too many indices for array: array is 3-dimensional, but 4 were indexed

复现环境 Environment

-OS: windows -PaddlePaddle: 2.6.0 -PaddleSeg: release/2.9 -python: 3.8.10 -Cuda: 11.8 -cudnn: 8.9.7

Bug描述确认 Bug description confirmation

• [X] 我确认已经提供了Bug复现步骤、代码改动说明、以及环境信息，确认问题是可以复现的。I confirm that the bug replication steps, code change instructions, and environment information have been provided, and the problem can be reproduced.

是否愿意提交PR？ Are you willing to submit a PR?

• [X] 我愿意提交PR！I'd like to help by submitting a PR!

[andy315]

export 的时候带上output_op参数（试了softmax）没有问题。default 是argmax，难道这个有问题？

[shiyutang]

看上去是预测图像中的问题，我们会尽快确认排除。

[chenjjcccc]

export 的时候带上output_op参数（试了softmax）没有问题。default 是argmax，难道这个有问题？

我用提供的推理模型测试好像是没问题的

[tongchangD]

@andy315 我 也遇到了相似的问题 ，如果是用download_pretrained_models.py 这个代码下载下来的模型是能用的，但是简介里提供的其他模型是用不了的。不知为何 请问你知道问题出在哪里不？

[tongchangD]

突然发现是自己蠢了，提供了 argmax 和 softmax的 ，我不该用softmax的代码 argmax的模型