GCCN: Global Context Convolutional Network

TL;DR

GCCN introduces a global context feature augmentation method for visual recognition, significantly improving accuracy on standard datasets and few-shot learning benchmarks by leveraging local maxima in image patches.

Contribution

The paper presents GCCN, a novel approach that incorporates global contextual features into convolutional networks, enhancing performance in image classification and few-shot learning tasks.

Findings

Achieves 94.6% on CIFAR-10 and 95.41% on STL-10.

Outperforms state-of-the-art few-shot methods on Omniglot, MiniImageNet, and CUB-200.

Improves accuracy of prototypical and matching networks by up to 30%.

Abstract

In this paper, we propose Global Context Convolutional Network (GCCN) for visual recognition. GCCN computes global features representing contextual information across image patches. These global contextual features are defined as local maxima pixels with high visual sharpness in each patch. These features are then concatenated and utilised to augment the convolutional features. The learnt feature vector is normalised using the global context features using Frobenius norm. This straightforward approach achieves high accuracy in compassion to the state-of-the-art methods with 94.6% and 95.41% on CIFAR-10 and STL-10 datasets, respectively. To explore potential impact of GCCN on other visual representation tasks, we implemented GCCN as a based model to few-shot image classification. We learn metric distances between the augmented feature vectors and their prototypes representations, similar to Prototypical and Matching Networks. GCCN outperforms state-of-the-art few-shot learning methods achieving 99.9%, 84.8% and 80.74% on Omniglot, MiniImageNet and CUB-200, respectively. GCCN has significantly improved on the accuracy of state-of-the-art prototypical and matching networks by up to 30% in different few-shot learning scenarios.