Chunfeng Song

大部分mask质量是理想的，仅有少部分是分割失败的。这些不理想的情况多是由于遮挡、原始图像残缺等原因导致。在我们的这个工作中，没有舍弃这些数据，而是跟原始图像一样正常使用了。理论上若有比较好的滤除差样本的机制，效果应该会更好。

We get the foreground image with a well-trained FCN model. We are sorry that we could not share this segmentation model. You can refer to the paper of "Early Hierarchical...

Hi, there are some differences with above two paper. 1) We re-design and modify the FCN based model motivatied by above two papers. However, the model used in our CVPR...

Sorry, we can't share the segmentation data. I think one possible solution is that, you can try to select some released high-quality masks and its RGB images of Market-1501 and...

Our pytorch version is not ready now. I think you can refer to this project now: [reid-mask](https://github.com/ChengpengChen/Reid-mask)

Hi, the model weights are in '/experiments/', e.g., '/experiments/mars/' for models training on MARS dataset.

You may refer to this issiue #1.

@prashant-bansod Hi, this project is just for demo, which is very easy for new learner. If you want to learn gait recognition in depth, you can follow the SOTA gaitset...

Thanks for your interest! As the original code is in CAFFE, which is not very popular now. We are trying to re-implement our model in PyTorch. The PyTorch code will...

@zhouyongxiu Hi! The CAFFE code is released now. We are sorry that the PyTorch version is still not ready.