Fast Real-Time Pipeline for Robust Arm Gesture Recognition
This work addresses gesture recognition for traffic control applications, but it is incremental as it builds on existing methods like OpenPose and RNNs with minor enhancements.
The paper tackles real-time arm gesture recognition by developing a pipeline using OpenPose keypoints, normalization, and an RNN classifier, achieving high accuracy on a custom traffic-control dataset across varying viewing angles and speeds.
This paper presents a real-time pipeline for dynamic arm gesture recognition based on OpenPose keypoint estimation, keypoint normalization, and a recurrent neural network classifier. The 1 x 1 normalization scheme and two feature representations (coordinate- and angle-based) are presented for the pipeline. In addition, an efficient method to improve robustness against camera angle variations is also introduced by using artificially rotated training data. Experiments on a custom traffic-control gesture dataset demonstrate high accuracy across varying viewing angles and speeds. Finally, an approach to calculate the speed of the arm signal (if necessary) is also presented.