Kevinz-code

> The later is what I mean, thank you for pointing out the confusion. Get Outlook for Android > […](#) > ________________________________ From: kevin655 Sent: Tuesday, December 8, 2020 6:25:32...

Hi, thanks for reading our work. A core implementation of the heatmap's visualization has been added in our codebase, see utils/visualize.py. This file demonstrate a simple visualization of an image...

Hi, thank you for reproducing our work and the baseline methods. When we run baseline method on COCO, the learning rate and other configs follow previous multi-label work, e.g. [ML-GCN](https://openaccess.thecvf.com/content_CVPR_2019/papers/Chen_Multi-Label_Image_Recognition_With_Graph_Convolutional_Networks_CVPR_2019_paper.pdf),...

Hi, @ghylander Thanks for your question and implementation. Here are three key steps in CSRA module: 1. generate the attention score s^i_j and class specific feature a^i (Eq. 2 and...

Hi, thanks for reading. The 'main.py' in our codebase describes the setting of wider attribute dataset using VIT-CSRA models, say, step_size=5, lr_backbone=0.1, lr_fc=0.01. You can follow the same settings when...

Hi, @sure7018, I'm not sure I fully understand your question. You mean the improvement of the VIT structure, or the improvement of our VIT-CSRA model compared with orginal VIT model?...

Hi, thanks for reading. 1. We use "Normalization" because it could lead to faster convergence (less training time) in our experiment, but it usually has no impact on the final...

Hi, @yangt1013 Actually the ""def val()"" in main.py and the "def val()" in val.py are doing the same function. For example, in VOC2007, these two "def val()" are validating the...

Hi, Thanks for reading. Actually, partial-label dataset are those with a candidate label set for each image, such as CIFAR-partial label dataset. These datasets usually contain one object class in...

Hi, Thanks for reading. We did have the models you referred to and we can release them in one or two weeks. Best, Ke Zhu.