Ma Shuhao comments

Results 20 comments of


                                            Ma Shuhao

hello,where is the utils?

This code is not a complete version. I had seen a complete version on github ,but I forget that link.

Is there some question in transform.py?

Sorry,this code have a trouble def __call__(self, gray_image): size = gray_image.size() print(gray_image) color_image = torch.ByteTensor(3, size[0], size[1]).fill_(0) - for label in range(1, len(self.cmap)): mask = gray_image[0] == label **color_image[0][mask] =...

感觉这份代码对于实际工作几乎没什么意义

也就是说当提取了实例信息，那么每个box中的点云属于哪个类别并不清楚

感觉这份代码对于实际工作几乎没什么意义

这工作还不如直接使用pointnet++获取语义信息，再用BFS提取实例

你好。我训练COCO数据集，发现先关键点loss居高不下

您好，我发现COCO数据集收敛了。但是我自己用labelme标注了2张图片，其中有人的bounding box和3个keypoints 但是不收敛，好难过，方便吗，大佬，加一下qq

你好。我训练COCO数据集，发现先关键点loss居高不下

大佬，我QQ是370308707

I think there is a trouble code

@dronefreak I have use the other version code of bisenet

I think there is a trouble code

@mshmoon The CamVid training codes can be any semantic segmenation codes, but the difference is the label encoding and label decoding betweent these codes, do you have a trouble about...

I think there is a trouble code

@mshmoon I have train CamVid, Cityscapes and Pascal VOC2012