How to predict one response map for each vehicle key point

Zhongdao / VehicleReIDKeyPointData

Annotations of key point location and vehicle orientation for VeRi-776 dataset. ICCV'17 paper: Orientation Invariant Feature Embedding and Spatial Temporal Regularization for Vehicle Re-identification.

MIT License

158 stars 23 forks source link

How to predict one response map for each vehicle key point #1

Open pengw416 opened 5 years ago

pengw416 commented 5 years ago

请问在代码当中是如何从一张训练集图片预测得到含有20个关键点的车辆响应图？

pengw416 commented 5 years ago

keypoint_test.txt和keypoint_train.txt是如何得到的？

Zhongdao commented 5 years ago

The key points provided in this repo are manually annotated. We then use the annotations as supervision to train an Hourglass-like Network[1] for automatically predicting the key points, given an input vehicle image.

[1].A. Newell, K. Yang, and J. Deng. Stacked hourglass networks for human pose estimation. In ECCV, 2016