Can Discriminative Cues Aid Face Recognition Across Age?

Can significant gains in face recognition performance can be achieved with the aid of cognitive evidences like gender and age group of a person?. This work focuses on studying the impact of applying these discriminative cues to a face recognition system in its performance. We propose a pipelined framework which includes a classification system for extracting the gender and age group information of a given image. These face recognition system at the end of the pipeline uses these information to select a subset of the search space to perform recognition. Experimental results demonstrate that categorizing the images based on discriminative cues offers significant improvements in recognition performance including accuracy, lower time requirements, and graceful degradation.

Proposed Pipeline

The proposed pipeline includes a Gender and Age classifier whose classification results are used by the face recognition system in performing the recognition. The pipeline is given below.

Proposed Pipeline for Gender and Age aided FR

Feature Extraction and Classification

The facial features are extracted using the Local Gabor Binary Patterns which is a combination of the Gabor features and the Local Binary patterns. Random forest is used for classification purposes. The feature extraction process is explained in the figure below.

Face Feature Extraction using LGBP

Related Publications

G. Mahalingam and C. KAmbhamettu, Can Discriminative Cues Aid Face Recognition Across Age?, To Appear in The 9th IEEE Conference on Automatic Face and Gesture Recognition, 2011.

Big View Video/Image Modeling and Synthesis Lab Home
VIMS Home People Research Publications Workshops Achievements Courses Software Data Photos login	Can Discriminative Cues Aid Face Recognition Across Age? Can significant gains in face recognition performance can be achieved with the aid of cognitive evidences like gender and age group of a person?. This work focuses on studying the impact of applying these discriminative cues to a face recognition system in its performance. We propose a pipelined framework which includes a classification system for extracting the gender and age group information of a given image. These face recognition system at the end of the pipeline uses these information to select a subset of the search space to perform recognition. Experimental results demonstrate that categorizing the images based on discriminative cues offers significant improvements in recognition performance including accuracy, lower time requirements, and graceful degradation. Proposed Pipeline The proposed pipeline includes a Gender and Age classifier whose classification results are used by the face recognition system in performing the recognition. The pipeline is given below. Proposed Pipeline for Gender and Age aided FR Feature Extraction and Classification The facial features are extracted using the Local Gabor Binary Patterns which is a combination of the Gabor features and the Local Binary patterns. Random forest is used for classification purposes. The feature extraction process is explained in the figure below. Face Feature Extraction using LGBP Related Publications G. Mahalingam and C. KAmbhamettu, Can Discriminative Cues Aid Face Recognition Across Age?, To Appear in The 9th IEEE Conference on Automatic Face and Gesture Recognition, 2011. Page last modified on January 16, 2011, at 05:36 PM