Directing DNNs Attention for Facial Attribution Classification using Gradient-weighted Class Activation Mapping

TL;DR

This paper introduces an interactive method to guide DNNs to focus on relevant facial regions using Grad-CAM, reducing co-occurrence bias and improving attribute classification transferability.

Contribution

It presents a novel approach to direct neural network attention interactively, enhancing facial attribute classification by mitigating bias.

Findings

Improved focus on relevant facial regions with Grad-CAM

Reduced co-occurrence bias in attribute classification

Enhanced transferability of pre-trained DNNs

Abstract

Deep neural networks (DNNs) have a high accuracy on image classification tasks. However, DNNs trained by such dataset with co-occurrence bias may rely on wrong features while making decisions for classification. It will greatly affect the transferability of pre-trained DNNs. In this paper, we propose an interactive method to direct classifiers paying attentions to the regions that are manually specified by the users, in order to mitigate the influence of co-occurrence bias. We test on CelebA dataset, the pre-trained AlexNet is fine-tuned to focus on the specific facial attributes based on the results of Grad-CAM.