Robustness of Object Recognition under Extreme Occlusion in Humans and Computational Models

TL;DR

This study compares human and computational object recognition under extreme occlusion, revealing humans' robustness, CNNs' limitations, and the effectiveness of compositional models in handling real-world occlusion.

Contribution

It demonstrates that current CNNs lack robustness to real-world occlusion and introduces a compositional model that performs better, aligning more closely with human recognition patterns.

Findings

Humans are highly robust to extreme occlusion.

Standard CNNs perform poorly under extreme occlusion.

Compositional models outperform CNNs and resemble human error patterns.

Abstract

Most objects in the visual world are partially occluded, but humans can recognize them without difficulty. However, it remains unknown whether object recognition models like convolutional neural networks (CNNs) can handle real-world occlusion. It is also a question whether efforts to make these models robust to constant mask occlusion are effective for real-world occlusion. We test both humans and the above-mentioned computational models in a challenging task of object recognition under extreme occlusion, where target objects are heavily occluded by irrelevant real objects in real backgrounds. Our results show that human vision is very robust to extreme occlusion while CNNs are not, even with modifications to handle constant mask occlusion. This implies that the ability to handle constant mask occlusion does not entail robustness to real-world occlusion. As a comparison, we propose another computational model that utilizes object parts/subparts in a compositional manner to build robustness to occlusion. This performs significantly better than CNN-based models on our task with error patterns similar to humans. These findings suggest that testing under extreme occlusion can better reveal the robustness of visual recognition, and that the principle of composition can encourage such robustness.