AL-GTD: Deep Active Learning for Gaze Target Detection

TL;DR

AL-GTD introduces an active learning approach for gaze target detection that significantly reduces the need for large labeled datasets by selecting the most informative samples, achieving high accuracy with less data.

Contribution

It presents a novel active learning framework combining supervised and self-supervised losses with pseudo-labeling for gaze target detection, reducing data annotation efforts.

Findings

Achieves top AUC results with only 40-50% of training data.

Reaches satisfactory performance with 10-20% of training data.

Outperforms existing active learning methods and state-of-the-art detectors in low-data regimes.

Abstract

Gaze target detection aims at determining the image location where a person is looking. While existing studies have made significant progress in this area by regressing accurate gaze heatmaps, these achievements have largely relied on access to extensive labeled datasets, which demands substantial human labor. In this paper, our goal is to reduce the reliance on the size of labeled training data for gaze target detection. To achieve this, we propose AL-GTD, an innovative approach that integrates supervised and self-supervised losses within a novel sample acquisition function to perform active learning (AL). Additionally, it utilizes pseudo-labeling to mitigate distribution shifts during the training phase. AL-GTD achieves the best of all AUC results by utilizing only 40-50% of the training data, in contrast to state-of-the-art (SOTA) gaze target detectors requiring the entire training dataset to achieve the same performance. Importantly, AL-GTD quickly reaches satisfactory performance with 10-20% of the training data, showing the effectiveness of our acquisition function, which is able to acquire the most informative samples. We provide a comprehensive experimental analysis by adapting several AL methods for the task. AL-GTD outperforms AL competitors, simultaneously exhibiting superior performance compared to SOTA gaze target detectors when all are trained within a low-data regime. Code is available at https://github.com/francescotonini/al-gtd.