Black-Box Optimization of Object Detector Scales

TL;DR

This paper explores automatic hyper-parameter tuning for object detectors using black-box optimization methods, leading to notable improvements in detection accuracy on standard datasets.

Contribution

It introduces the application of Bayesian Optimization, SMAC, and CMA-ES to tune object detector hyper-parameters, specifically prior box scales, which was previously manual.

Findings

mAP increases by 2% on PASCAL VOC 2007 with Faster R-CNN

mAP increases by 3% on PASCAL VOC 2007 with SSD

medium and large object detection improves on COCO dataset

Abstract

Object detectors have improved considerably in the last years by using advanced CNN architectures. However, many detector hyper-parameters are generally manually tuned, or they are used with values set by the detector authors. Automatic Hyper-parameter optimization has not been explored in improving CNN-based object detectors hyper-parameters. In this work, we propose the use of Black-box optimization methods to tune the prior/default box scales in Faster R-CNN and SSD, using Bayesian Optimization, SMAC, and CMA-ES. We show that by tuning the input image size and prior box anchor scale on Faster R-CNN mAP increases by 2% on PASCAL VOC 2007, and by 3% with SSD. On the COCO dataset with SSD there are mAP improvement in the medium and large objects, but mAP decreases by 1% in small objects. We also perform a regression analysis to find the significant hyper-parameters to tune.