Human Action Recognition Algorithm Based on Improved ResNet and Skeletal Keypoints in Single Image

Figure 11

Keypoint detection renderings in different scenes, (a) single person with no occlusion, (b) multiple people with occlusion, (c) small size of multiple people with occlusion, (d) half body without occlusion, (e) multiple people without occlusion, and (f) small size of single person without occlusion.

(a)

(b)

(c)

(d)

(e)

(f)