Generative Modeling for Small-Data Object Detection

TL;DR

This paper introduces a novel generative modeling approach with an unrolling mechanism to improve small-data object detection, significantly enhancing detection accuracy in medical and pedestrian datasets.

Contribution

It proposes a new generative model with an unrolling mechanism that jointly optimizes image generation and detection, outperforming existing methods in small-data scenarios.

Findings

Improves average precision on NIH Chest X-ray by 20%.

Enhances localization accuracy by 50%.

Outperforms state-of-the-art on two challenging datasets.

Abstract

This paper explores object detection in the small data regime, where only a limited number of annotated bounding boxes are available due to data rarity and annotation expense. This is a common challenge today with machine learning being applied to many new tasks where obtaining training data is more challenging, e.g. in medical images with rare diseases that doctors sometimes only see once in their life-time. In this work we explore this problem from a generative modeling perspective by learning to generate new images with associated bounding boxes, and using these for training an object detector. We show that simply training previously proposed generative models does not yield satisfactory performance due to them optimizing for image realism rather than object detection accuracy. To this end we develop a new model with a novel unrolling mechanism that jointly optimizes the generative model and a detector such that the generated images improve the performance of the detector. We show this method outperforms the state of the art on two challenging datasets, disease detection and small data pedestrian detection, improving the average precision on NIH Chest X-ray by a relative 20% and localization accuracy by a relative 50%.