We’ve had a publication accepted at ECCV 2018! Here’s the abstract:
This work presents a novel hand pose estimation framework via intermediate dense guidance map supervision. By leveraging the advantage of predicting heat maps of hand joints in detection-based methods, we propose to use dense feature maps through intermediate supervision in a regression-based framework that is not limited to the resolution of the heat map. Our dense feature maps are delicately designed to encode the hand geometry and the spatial relation between local joint and global hand. The proposed framework significantly improves the state-of-the-art in both 2D and 3D on the recent benchmark datasets.
Xiaokun Wu wrote up an awesome post on the project on his website. I recommend reading it! They say pictures speak a thousand words, so here’s a gif Xiaokun made giving the general idea in a succint way.
Wu, X., Finnegan, D. J., O’Neill, E., Yang, Y.
HandMap: Robust Hand Pose Estimation via Intermediate Dense Guidance Map Supervision.
ECCV 2018: Proceedings of the 15th European Conference on Computer Vision.