AttaNet icon indicating copy to clipboard operation
AttaNet copied to clipboard

AttaNet for real-time semantic segmentation.

Results 5 AttaNet issues
Sort by recently updated
recently updated
newest added

根据如下几个issue: https://github.com/songqi-github/AttaNet/issues/16 https://github.com/songqi-github/AttaNet/issues/13 **这篇论文的Teaser结果完完全全是错误的。** 实时模型不能用多尺度crop测试结果却报告小图分辨率的速度。 全图测试结果attanet就76不到的水平。 另外速度测试没有用cuda的同步命令,速度测试也不对。

作者你好,关于速度测试我有一些疑问,能不能认为你在测速的时候先把图片放缩到更小的resolution,如512x1024,然后把放缩后的图片输入到模型里面去,再将输出放缩回1024x2048?如果真是如此的话,那你此时用的测速resolution不能声称为1024x2048,而是512x1024。要不你就测精度的时候相应地如此操作,否则无法确保公平。

When I was looking at the SAM part of the code and the paper, I found that the operations of Q and K do not correspond to each other. If...

Is there has a python file for the ade20k dataset except for the cityscapes?