View-invariant gait recognition system using a gait energy image decomposition method

What is it about?

Recognition in surveillance environment is difficult due to the lack of active user participation. Gait is a bio-metric trait that allows non-intrusive and non-contact recognition. However gait in surveillance environment faces additional challenges such as view and appearance change. The paper presents a novel system to address these challenges.

Featured Image

Why is it important?

Gait recognition systems can be effectively performed in a controlled environment where the viewpoint of the camera and the appearance of the user is fixed. However the performance of the system degrades significantly with change in view or appearance. The paper presents a complete system that tackles change in view and change in appearance using only 2D-video sequences as input. The system effectively identifies the correct view and the altered parts of the user to obtain results that beat the state-of-the-art.