A biological vision inspired framework for machine perception of abutting grating illusory contours
This addresses a specific perception discrepancy for AI systems aiming for human-level intelligence, representing an incremental advance in aligning machine vision with human perception.
The paper tackles the problem that deep neural networks fail to perceive abutting grating illusory contours, which misaligns with human perception, by proposing a novel deep network called ICPNet inspired by visual cortex circuits, resulting in significant improvements in top-1 accuracy on test sets like AG-MNIST and AG-Fashion-MNIST.
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.