A biological vision inspired framework for machine perception of abutting grating illusory contours

TL;DR

This paper introduces ICPNet, a biologically inspired deep neural network designed to better perceive illusory contours like abutting gratings, aligning machine perception more closely with human visual cognition.

Contribution

The paper presents a novel deep network architecture with modules inspired by visual cortex circuits, improving DNN sensitivity to illusory contours compared to existing models.

Findings

ICPNet outperforms state-of-the-art models in detecting abutting grating illusions.

Significant improvements in top-1 accuracy on AG-MNIST and AG-Fashion-MNIST datasets.

The model demonstrates enhanced alignment with human perception of illusory contours.

Abstract

Higher levels of machine intelligence demand alignment with human perception and cognition. Deep neural networks (DNN) dominated machine intelligence have demonstrated exceptional performance across various real-world tasks. Nevertheless, recent evidence suggests that DNNs fail to perceive illusory contours like the abutting grating, a discrepancy that misaligns with human perception patterns. Departing from previous works, we propose a novel deep network called illusory contour perception network (ICPNet) inspired by the circuits of the visual cortex. In ICPNet, a multi-scale feature projection (MFP) module is designed to extract multi-scale representations. To boost the interaction between feedforward and feedback features, a feature interaction attention module (FIAM) is introduced. Moreover, drawing inspiration from the shape bias observed in human perception, an edge detection task conducted via the edge fusion module (EFM) injects shape constraints that guide the network to concentrate on the foreground. We assess our method on the existing AG-MNIST test set and the AG-Fashion-MNIST test sets constructed by this work. Comprehensive experimental results reveal that ICPNet is significantly more sensitive to abutting grating illusory contours than state-of-the-art models, with notable improvements in top-1 accuracy across various subsets. This work is expected to make a step towards human-level intelligence for DNN-based models.