I am working on gesture recognition in images and the best way that I am aware of, is whether using end to end approaches with deep neural networks or extracting body joint positions in an image and then doing some pre-processing on these joints positions and finally, feeding them to a classifier. I wanted to now what is the state of the art method for full body gesture recognition in images and if there are some good papers or a blog post reviewing different methods, as I have problem finding one.
1 Answer
If you are planing to train a model which is using images taken by a single camera, I encourage you to take a look at this paper. They use a camera to track a person and recognize gestures involving arm motion. In this paper, two alternative methods for gesture recognition are compared: a template based approach and a neural network approach.
However, another way which I suggest is using depth cameras such as Intel Realsense or Microsoft Kinect cameras. This paper uses depth cameras for full body gesture recognition. I think using these cameras take your project a step ahead and more accurate in initial states.