Bayesian Joint Topic Modelling for Weakly Supervised Object Localisation

TL;DR

This paper introduces a Bayesian joint topic model for weakly supervised object localisation, effectively leveraging background and unlabelled data to improve accuracy over previous methods.

Contribution

It presents a novel Bayesian framework that models all object classes and backgrounds jointly, enabling better disambiguation and integration of prior knowledge.

Findings

Outperforms state-of-the-art on VOC dataset

Effectively uses unlabelled data for learning

Improves localisation accuracy

Abstract

We address the problem of localisation of objects as bounding boxes in images with weak labels. This weakly supervised object localisation problem has been tackled in the past using discriminative models where each object class is localised independently from other classes. We propose a novel framework based on Bayesian joint topic modelling. Our framework has three distinctive advantages over previous works: (1) All object classes and image backgrounds are modelled jointly together in a single generative model so that "explaining away" inference can resolve ambiguity and lead to better learning and localisation. (2) The Bayesian formulation of the model enables easy integration of prior knowledge about object appearance to compensate for limited supervision. (3) Our model can be learned with a mixture of weakly labelled and unlabelled data, allowing the large volume of unlabelled images on the Internet to be exploited for learning. Extensive experiments on the challenging VOC dataset demonstrate that our approach outperforms the state-of-the-art competitors.