Auxiliary Bounding Box Regression for Object Detection in Optical Remote Sensing Imagery

摘要

Object detection in optical remote sensing imagery is being explored to deal witharbitrary orientations and complex appearance which is still a major issue in recentyears. To perceive a better solution to the addressed problem, the post-processingof bounding boxes (BBs) has been evaluated and discussed for the applications ofobject detection. In this paper, the proposed method has divided into two stages; thefirst stage is based on thresholding of BBs with respect to the confidence values andthe second stage is based on the area-based BB regression (BBR). In BBR, the areaof each BB was estimated then the oversized and undersized BBs were removed withrespect to the size of objects which are being detected. The widely known regionbasedapproaches RCNN, Fast-RCNN and Faster-RCNN are used for evaluation andcomparative analysis validates the proposed framework. The results show that theproposed post-processing is very effective for each kind of region-based detector.