Transferable Feature Representation for Visible-to-Infrared Cross-Dataset Human Action Recognition
This work addresses infrared human action recognition, which is important for applications like surveillance in low-light conditions, but it is incremental as it builds on existing transfer learning and domain adaptation methods.
The authors tackled the problem of limited infrared action data by proposing a transfer learning framework that uses auxiliary visible light data, achieving state-of-the-art performance on the InfAR dataset.
Recently, infrared human action recognition has attracted increasing attention for it has many advantages over visible light, that is, being robust to illumination change and shadows. However, the infrared action data is limited until now, which degrades the performance of infrared action recognition. Motivated by the idea of transfer learning, an infrared human action recognition framework using auxiliary data from visible light is proposed to solve the problem of limited infrared action data. In the proposed framework, we first construct a novel Cross-Dataset Feature Alignment and Generalization (CDFAG) framework to map the infrared data and visible light data into a common feature space, where Kernel Manifold Alignment (KEMA) and a dual alignedto-generalized encoders (AGE) model are employed to represent the feature. Then, a support vector machine (SVM) is trained, using both the infrared data and visible light data, and can classify the features derived from infrared data. The proposed method is evaluated on InfAR, which is a publicly available infrared human action dataset. To build up auxiliary data, we set up a novel visible light action dataset XD145. Experimental results show that the proposed method can achieve state-of-the-art performance compared with several transfer learning and domain adaptation methods.