An approach to partial occlusion using deep metric learning

<span>The human face can be used as an identification and authentication tool in biometric systems. Face recognition forensics is a challenging task due to the presence of partial occlusion features like wearing hat, sunglasses, scarf, beard. In forensics, criminal having most difficult perform. this paper, combination histogram gradients (HOG) with Euclidean distance proposed. Deep metric learning process measuring similarity between samples using optimal metrics for tasks. proposed system, deep technique HOG generate 128d real feature vector. then applied vectors tolerance threshold set decide whether it match or mismatch. Experiments are carried out on disguised faces wild (DFW) dataset collected from IIIT Delhi which consists 1000 subjects 600 were testing remaining 400 training purposes. The system provides accuracy 89.8% outperforms compared other existing methods.</span>