The Ikshana Hypothesis of Human Scene Understanding

TL;DR

This paper introduces the Ikshana hypothesis inspired by neuroscience to improve deep neural networks for scene understanding, leading to a novel CNN architecture that outperforms existing methods on semantic segmentation benchmarks.

Contribution

The paper proposes the Ikshana hypothesis as a neuroscience-inspired perspective for designing neural networks, resulting in the IkshanaNet architecture for improved scene understanding.

Findings

IkshanaNet outperforms baselines on Cityscapes and CamVid benchmarks.

The hypothesis links neuroscience insights to CNN design.

Empirical validation shows improved semantic segmentation results.

Abstract

In recent years, deep neural networks (DNNs) achieved state-of-the-art performance on several computer vision tasks. However, the one typical drawback of these DNNs is the requirement of massive labeled data. Even though few-shot learning methods address this problem, they often use techniques such as meta-learning and metric-learning on top of the existing methods. In this work, we address this problem from a neuroscience perspective by proposing a hypothesis named Ikshana, which is supported by several findings in neuroscience. Our hypothesis approximates the refining process of conceptual gist in the human brain while understanding a natural scene/image. While our hypothesis holds no particular novelty in neuroscience, it provides a novel perspective for designing DNNs for vision tasks. By following the Ikshana hypothesis, we design a novel neural-inspired CNN architecture named IkshanaNet. The empirical results demonstrate the effectiveness of our method by outperforming several baselines on the entire and subsets of the Cityscapes and the CamVid semantic segmentation benchmarks.