no code implementations • European Conference on Computer Vision 2019 • Jingxuan Hou
In this paper, we propose an end-to-end Spatial-Temporal Attention Residual Temporal Convolutional Network (STA-Res-TCN) for skeleton-based dynamic hand gesture recognition, which learns different levels of attention and assigns them to each spatial-temporal feature extracted by the convolution filters at each time step.
no code implementations • 22 Nov 2017 • Jingxuan Hou, Tae Soo Kim, Austin Reiter
Based on the findings from the model interpretation analysis, we propose a targeted refinement technique, which can generalize to various other recognition models.