How To Effectively Train An Ensemble Of Faster R-CNN Object Detectors To Quantify Uncertainty

TL;DR

This paper introduces an efficient method for training ensemble Faster R-CNN models to estimate uncertainty in object detection, significantly reducing training time while maintaining performance.

Contribution

The paper proposes training a single RPN and multiple heads for ensemble uncertainty estimation, offering a faster alternative to fully training multiple models.

Findings

Ensemble models with shared RPN are faster to train.

The approach achieves comparable uncertainty estimation performance.

The method outperforms Gaussian YOLOv3 in certain metrics.

Abstract

This paper presents a new approach for training two-stage object detection ensemble models, more specifically, Faster R-CNN models to estimate uncertainty. We propose training one Region Proposal Network(RPN) and multiple Fast R-CNN prediction heads is all you need to build a robust deep ensemble network for estimating uncertainty in object detection. We present this approach and provide experiments to show that this approach is much faster than the naive method of fully training all $n$ models in an ensemble. We also estimate the uncertainty by measuring this ensemble model's Expected Calibration Error (ECE). We then further compare the performance of this model with that of Gaussian YOLOv3, a variant of YOLOv3 that models uncertainty using predicted bounding box coordinates. The source code is released at \url{https://github.com/Akola-Mbey-Denis/EfficientEnsemble}