We don't need no bounding-boxes: Training object class detectors using only human verification

TL;DR

This paper introduces a novel training scheme for object detectors that relies solely on human verification of automatically generated bounding-boxes, significantly reducing annotation time while maintaining high detection quality.

Contribution

The authors propose a verification-based training method that eliminates the need for manual bounding-box annotation, improving efficiency and nearly matching fully supervised detector performance.

Findings

High-quality detectors achieved with verification-based training

Detection performance close to fully supervised methods

Annotation time reduced by a factor of 6 to 9

Abstract

Training object class detectors typically requires a large set of images in which objects are annotated by bounding-boxes. However, manually drawing bounding-boxes is very time consuming. We propose a new scheme for training object detectors which only requires annotators to verify bounding-boxes produced automatically by the learning algorithm. Our scheme iterates between re-training the detector, re-localizing objects in the training images, and human verification. We use the verification signal both to improve re-training and to reduce the search space for re-localisation, which makes these steps different to what is normally done in a weakly supervised setting. Extensive experiments on PASCAL VOC 2007 show that (1) using human verification to update detectors and reduce the search space leads to the rapid production of high-quality bounding-box annotations; (2) our scheme delivers detectors performing almost as good as those trained in a fully supervised setting, without ever drawing any bounding-box; (3) as the verification task is very quick, our scheme substantially reduces total annotation time by a factor 6x-9x.