Thermal Image Degradation Influence on R-CNN Face Detection Performance

Abstract

Visible light face detection systems have been well researched and in controlled environments can reach excellent accuracy. Variation in illumination conditions results in performance degradation and illumination is the one of the major limitations in visible light face detection systems. Using thermal infrared cameras one can provide a solution to this problem. Recent studies show that deep learning approaches can achieve impressive performance on object detection tasks, and face detection in particular. The goal of this paper is to find an effective way to take advantages from thermal IR spectra and provide a comparative analysis of various image degradation influence on thermal face detection performance in a system based on R-CNN.