From Volcano to Toyshop: Adaptive Discriminative Region Discovery for Scene Recognition

TL;DR

This paper introduces Adi-Red, an adaptive method for scene recognition that discovers discriminative regions using a pre-trained CNN, outperforming existing methods by effectively focusing on scene-specific features.

Contribution

Adi-Red is the first adaptive discriminative region discovery approach tailored for scene recognition, leveraging scene-specific properties for improved accuracy.

Findings

Outperforms state-of-the-art on SUN397 dataset

Demonstrates advantages on Places dataset

Effectively avoids noise while capturing important scene information

Abstract

As deep learning approaches to scene recognition emerge, they have continued to leverage discriminative regions at multiple scales, building on practices established by conventional image classification research. However, approaches remain largely generic, and do not carefully consider the special properties of scenes. In this paper, inspired by the intuitive differences between scenes and objects, we propose Adi-Red, an adaptive approach to discriminative region discovery for scene recognition. Adi-Red uses a CNN classifier, which was pre-trained using only image-level scene labels, to discover discriminative image regions directly. These regions are then used as a source of features to perform scene recognition. The use of the CNN classifier makes it possible to adapt the number of discriminative regions per image using a simple, yet elegant, threshold, at relatively low computational cost. Experimental results on the scene recognition benchmark dataset SUN397 demonstrate the ability of Adi-Red to outperform the state of the art. Additional experimental analysis on the Places dataset reveals the advantages of Adi-Red, and highlight how they are specific to scenes. We attribute the effectiveness of Adi-Red to the ability of adaptive region discovery to avoid introducing noise, while also not missing out on important information.