PROJECT TITLE :

Joint Hand Detection and Rotation Estimation Using CNN - 2018

ABSTRACT:

Hand detection is important for several hand related tasks, e.g., recovering hand cause and understanding gesture. However, hand detection in uncontrolled environments is challenging because of the pliability of wrist joint and cluttered background. We have a tendency to propose a convolutional neural network (CNN), that formulates in-plane rotation explicitly to unravel hand detection and rotation estimation jointly. Our network architecture adopts the backbone of faster R-CNN to generate rectangular region proposals and extract local options. The rotation network takes the feature as input and estimates an in-plane rotation that manages to align the hand, if any in the proposal, to the upward direction. A derotation layer is then designed to explicitly rotate the native spatial feature map in step with the rotation network and feed aligned feature map for detection. Experiments show that our method outperforms the state-of-the-art detection models on widely-used benchmarks, such as Oxford and Egohands database. More analysis show that rotation estimation and classification will mutually benefit every alternative.