Yuqing Wang
@xllau Have you modified the post-processing file based on the inference instruction in README?
> Hello, good question! I've also faced this problem before. You can try this one: > #distribute model on first 7 gpus > model = torch.nn.DataParallel(model, device_ids=[0,1,2,3,4,5,6]) > images =...
> 不是，我没具体看过yolact论文，就是知道它的一个大概思想。 > 我的实现处理和那篇论文不一样，应该。 环境配置和运行方式可否说明一下，这两天有空跑一下,在coco上的AP还没有是吗
> @Epiphqny 配置，运行方式 与readme里的centernet代码 一样，把相应的py放入对应文件夹就行了 what's the meaning of mask_small, mask_mid and mask_large? outputs of FPN?
> @Epiphqny > 那些没有用到。那些是基于tensormask的一种方法，我训不出来，效果不好。 看不太懂你的图，加个微信？1600975602
> @Epiphqny > 那些没有用到。那些是基于tensormask的一种方法，我训不出来，效果不好。 我之前有基于yolact的思想在centernet上面做过，但是效果一般，和polarmask差不多的精度
@abubakar2141731 In google it seems like a torch version issue
@JRGit4UE It seems that the mmcv version is not correct.
Hi @LiWentomng, please refer to the official implementation for this issue. https://github.com/WXinlong/SOLO.
@lxtGH I have not tried less number convolutions, but I think the convolution number in the middle layer is not very important, so you could try less number.