同时,针对动作识别任务,一个自然的演进就是在骨架关键点坐标的基础上引入视觉特征,如图像特征,乃至场景图(scene-graph)等,并将它们统一在 ST-GCN 的分析框架下。
approach and 19-683% over the state-of-the-art with comparable detection accuracy. 【3】 Spatio-Temporal Scene-Graph To address these limitations, we propose sg2vec, a spatio-temporal scene-graph embedding methodology