PROJECT TITLE :
Joint Feature Learning for Face Recognition
This paper presents a brand new joint feature learning (JFL) approach to automatically learn feature representation from raw pixels for face recognition. Unlike many existing face recognition systems, where typical feature descriptors, like native binary patterns and Gabor options, are used for face illustration, we tend to propose an unsupervised feature learning method to find out hierarchical feature illustration. Since different face regions have different physical characteristics, we have a tendency to propose to use different feature dictionaries to represent them, and to find out multiple nevertheless connected feature projection matrices for these regions simultaneously. Hence position-specific discriminative data will be exploited for face illustration. Having learned these feature projections for different face regions, we tend to perform spatial pooling for face patches among each region to boost the representative power of the learned options. Moreover, we have a tendency to stack our JFL model into a deep design to take advantage of hierarchical data for feature illustration and any improve the popularity performance. Experimental results on 5 widely used face information sets show the effectiveness of our proposed approach.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here