Learning to Separate: Detecting Heavily-Occluded Objects in Urban Scenes
It addresses a critical issue for autonomous driving and surveillance systems by improving object detection in crowded urban environments where occlusions are common.
The paper tackles the problem of detecting heavily-occluded objects in urban scenes by proposing a Non-Maximum-Suppression algorithm that improves detection recall while maintaining high precision, achieving state-of-the-art performance on KITTI and CityPersons datasets.
While visual object detection with deep learning has received much attention in the past decade, cases when heavy intra-class occlusions occur have not been studied thoroughly. In this work, we propose a Non-Maximum-Suppression (NMS) algorithm that dramatically improves the detection recall while maintaining high precision in scenes with heavy occlusions. Our NMS algorithm is derived from a novel embedding mechanism, in which the semantic and geometric features of the detected boxes are jointly exploited. The embedding makes it possible to determine whether two heavily-overlapping boxes belong to the same object in the physical world. Our approach is particularly useful for car detection and pedestrian detection in urban scenes where occlusions often happen. We show the effectiveness of our approach by creating a model called SG-Det (short for Semantics and Geometry Detection) and testing SG-Det on two widely-adopted datasets, KITTI and CityPersons for which it achieves state-of-the-art performance.