BiSeNet icon indicating copy to clipboard operation
BiSeNet copied to clipboard

Implementation of BiSeNet and BiSeNetV2

Results 6 BiSeNet issues
Sort by recently updated
recently updated
newest added

1. Would you be putting the official code for BiSeNet V2? 2. What is the license?

首先感谢大佬 @ycszen 的工作:) 文章中有些我觉得写的不是很清楚的地方想问一下,关于generalization to large models章节中提到的两个参数α和d。 ![Screenshot from 2020-04-16 19-52-47](https://user-images.githubusercontent.com/15725187/79453548-6dee6100-801c-11ea-880d-228b89cf8999.png) 关于α,我觉得指的是进入segmentation heads模块之前的channel expansion倍数,如下图所示 ![Screenshot from 2020-04-16 19-51-51](https://user-images.githubusercontent.com/15725187/79454246-72674980-801d-11ea-8dcf-f91c76135c8f.png) 那么这个d参数指的是什么呢,文章中说d参数控制的是模型的深度,那么这个d是指代上文中的什么参数呢,是module repeat的次数吗. 还有点疑惑就是table5下方的描述α控制模型的channel capacity而d控制layer nums.然后下文中说α是width multiplier,d是depth multiplier。请问这个地方上下文的表述是否一致。 ![Screenshot from 2020-04-16 20-02-19](https://user-images.githubusercontent.com/15725187/79454072-3338f880-801d-11ea-9703-afe6af04d468.png) 这两个参数该如何理解更为准确呢。还请大佬有时间不吝赐教...

Hi, in the paper, 150K, 10K, 20K iterations for the Cityscapes dataset, CamVid dataset, and COCO-Stuff datasets respectively.... but image number of COCO db is much larger than Cityscapes.. why...

In Fig 6 of your paper, left input of sum is a tensor with shape of HxWxC, and right input is H/4xW/4xC, how can they be added ? Do you...