Scale-Invariant Convolutional Neural Networks

TL;DR

The paper introduces SiCNN, a scale-invariant CNN architecture that captures multi-scale features efficiently, improving robustness to object scale variations without increasing model size.

Contribution

It proposes a multi-column SiCNN with shared filters across scales, enabling scale invariance without enlarging the model.

Findings

SiCNN effectively detects features at multiple scales.

The model shows strong robustness against object scale variations.

Experimental results outperform traditional CNNs on scale-variant tasks.

Abstract

Even though convolutional neural networks (CNN) has achieved near-human performance in various computer vision tasks, its ability to tolerate scale variations is limited. The popular practise is making the model bigger first, and then train it with data augmentation using extensive scale-jittering. In this paper, we propose a scaleinvariant convolutional neural network (SiCNN), a modeldesigned to incorporate multi-scale feature exaction and classification into the network structure. SiCNN uses a multi-column architecture, with each column focusing on a particular scale. Unlike previous multi-column strategies, these columns share the same set of filter parameters by a scale transformation among them. This design deals with scale variation without blowing up the model size. Experimental results show that SiCNN detects features at various scales, and the classification result exhibits strong robustness against object scale variations.