Advancing TDFN: Precise Fixation Point Generation Using Reconstruction Differences

TL;DR

This paper improves the TDFN model by using image reconstruction differences to guide fixation point generation, leading to more accurate fixations and better classification performance with fewer fixations.

Contribution

It introduces a novel training method for fixation point generation based on reconstruction differences, enhancing accuracy and efficiency of the TDFN.

Findings

Achieves highly accurate fixation points

Significantly improves classification accuracy

Reduces the number of fixations needed

Abstract

Wang and Wang (2025) proposed the Task-Driven Fixation Network (TDFN) based on the fixation mechanism, which leverages low-resolution information along with high-resolution details near fixation points to accomplish specific visual tasks. The model employs reinforcement learning to generate fixation points. However, training reinforcement learning models is challenging, particularly when aiming to generate pixel-level accurate fixation points on high-resolution images. This paper introduces an improved fixation point generation method by leveraging the difference between the reconstructed image and the input image to train the fixation point generator. This approach directs fixation points to areas with significant differences between the reconstructed and input images. Experimental results demonstrate that this method achieves highly accurate fixation points, significantly enhances the network's classification accuracy, and reduces the average number of required fixations to achieve a predefined accuracy level.