On learning to localize objects with minimal supervision

TL;DR

This paper introduces a novel weakly supervised object localization method that uses only image-level labels, combining submodular cover and latent SVM techniques, achieving significant performance improvements on PASCAL VOC 2007.

Contribution

It presents a new approach that effectively localizes objects with minimal supervision, reducing annotation costs and improving detection accuracy.

Findings

50% relative improvement in mean average precision on PASCAL VOC 2007

Combines submodular cover with latent SVM for effective localization

Operates with only image-level labels, reducing annotation effort

Abstract

Learning to localize objects with minimal supervision is an important problem in computer vision, since large fully annotated datasets are extremely costly to obtain. In this paper, we propose a new method that achieves this goal with only image-level labels of whether the objects are present or not. Our approach combines a discriminative submodular cover problem for automatically discovering a set of positive object windows with a smoothed latent SVM formulation. The latter allows us to leverage efficient quasi-Newton optimization techniques. Our experiments demonstrate that the proposed approach provides a 50% relative improvement in mean average precision over the current state-of-the-art on PASCAL VOC 2007 detection.