Forecasting Human Dynamics from Static Images
This addresses the challenge of predicting human motion from static images for applications in robotics and computer vision, but it is incremental as it builds on existing methods for pose estimation and sequence prediction.
This paper tackles the problem of forecasting future human body poses from a single static image, proposing the 3D-PFNet which integrates pose estimation and sequence prediction to generate 3D pose sequences, achieving competitive performance in 2D pose forecasting and 3D pose recovery.
This paper presents the first study on forecasting human dynamics from static images. The problem is to input a single RGB image and generate a sequence of upcoming human body poses in 3D. To address the problem, we propose the 3D Pose Forecasting Network (3D-PFNet). Our 3D-PFNet integrates recent advances on single-image human pose estimation and sequence prediction, and converts the 2D predictions into 3D space. We train our 3D-PFNet using a three-step training strategy to leverage a diverse source of training data, including image and video based human pose datasets and 3D motion capture (MoCap) data. We demonstrate competitive performance of our 3D-PFNet on 2D pose forecasting and 3D pose recovery through quantitative and qualitative results.