2018/03/13 by Ayesha Gurnani, Kenil Shah, Gurnani, Ayesha +7
Computer Science · Psychology · #Computer Vision and Pattern Recognition (cs.CV) #Emotion and Mood Recognition #FOS: Computer and information sciences #Face and Expression Recognition #Face recognition and analysis
paper · pdf · doi:10.48550/arxiv.1803.05719
openalex publication_date 2018/03/13 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
How can we improve the facial soft-biometric classification with help of the\nhuman visual system? This paper explores the use of saliency which is\nequivalent to the human visual system to classify Age, Gender and Facial\nExpression soft-biometric for facial images. Using the Deep Multi-level Network\n(ML-Net) [1] and off-the-shelf face detector [2], we propose our approach -\nSAF-BAGE, which first detects the face in the test image, increases the\nBounding Box (B-Box) margin by 30%, finds the saliency map using ML-Net, with\n30% reweighted ratio of saliency map, it multiplies with the input cropped face\nand extracts the Convolutional Neural Networks (CNN) predictions on the\nmultiplied reweighted salient face. Our CNN uses the model AlexNet [3], which\nis pre-trained on ImageNet. The proposed approach surpasses the performance of\nother approaches, increasing the state-of-the-art by approximately 0.8% on the\nwidely-used Adience [28] dataset for Age and Gender classification and by\nnearly 3% on the recent AffectNet [36] dataset for Facial Expression\nclassification. We hope our simple, reproducible and effective approach will\nhelp ease future research in facial soft-biometric classification using\nsaliency.\n