Abstract:
To improve the efficiency in face recognition with highdimension features extracted from deep model, a fast recognition method based on hash coding is proposed. Different...Show MoreMetadata
Abstract:
To improve the efficiency in face recognition with highdimension features extracted from deep model, a fast recognition method based on hash coding is proposed. Different from others, the hash coding and the cascade network are designed for a two-stage face recognition. Firstly, the low-dimensional and high-dimensional features are extracted according to different models. Secondly, the low-dimensional features are quantized into hash codes by a piecewise function. And then, the first-identify is completed by calculating hamming distance between the hash codes. Finally, the second-identify is completed by calculating cosine distance between the high-dimensional features of face images after the first-identify. The experimental results show that the method proposed can improve the Rank-1 recognition efficiency up to 64% while the accuracy is the same as VGG.
Published in: 2017 13th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD)
Date of Conference: 29-31 July 2017
Date Added to IEEE Xplore: 25 June 2018
ISBN Information: