使用時序卷積和半監督訓練對視頻中的3D人類位姿估計（Facebook AI Research）

arxiv.org

1. 基於2D關鍵點的空洞時序卷積的全卷積模型。
2. 引入反投影，一個簡單有效的半監督方法，可以利用未標註的視頻數據。

摘要：

In this work, we demonstrate that 3D poses in video can be effectively estimated with a fully convolutional model based on dilated temporal convolutions over 2D key-points. We also introduce back-projection, a simple and effective semi-supervised training method that leverages unlabeled video data. We start with predicted 2D key-points for unlabeled video, then estimate 3D poses and finally back-project to the input 2D keypoints. In the supervised setting, our fully-convolutional model outper-forms the previous best result from the literature by 6 mm mean per-joint position error on Human3.6M, correspond-ing to an error reduction of 11%, and the model also shows significant improvements on HumanEva-I. More-over, experiments with back-projection show that it comfort-ably outperforms previous state-of-the-art results in semi-supervised settings where labeled data is scarce. Code and models are available at https://github.com/facebookresearch/VideoPose3D

效果動圖：

視頻：

https://www.zhihu.com/video/1051777230560100352
推薦閱讀：