How To Effectively Train An Ensemble Of Faster R-CNN Object Detectors To Quantify Uncertainty
This work addresses the need for efficient uncertainty quantification in object detection, which is incremental as it builds on existing Faster R-CNN and ensemble techniques.
The paper tackles the problem of efficiently training ensemble models for uncertainty estimation in object detection by proposing a method that trains one Region Proposal Network with multiple Fast R-CNN heads, resulting in much faster training than naive ensemble methods and providing uncertainty estimates via Expected Calibration Error.
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}