Local Region Proposing for Frame-Based Vehicle Detection in Satellite Videos

Abstract

Current new developments in remote sensing imagery enable satellites to capture videos from space. These satellite videos record the motion of vehicles over a vast territory, offering significant advantages in traffic monitoring systems over ground-based systems. However, detecting vehicles in satellite videos are challenged by the low spatial resolution and the low contrast in each video frame. The vehicles in these videos are small, and most of them are blurred into their background regions. While region proposals are often generated for efficient target detection, they have limited performance on satellite videos. To meet this challenge, we propose a Local Region Proposing approach (LRP) with three steps in this study. A video frame is segmented into semantic regions first and possible targets are then detected in these coarse scale regions. A discrete Histogram Mixture Model (HistMM) is proposed in the third step to narrow down the region proposals by quantifying their likelihoods towards the target category, where the training is conducted on positive samples only. Experiment results demonstrate that LRP generates region proposals with improved target recall rates. When a slim Fast-RCNN detector is applied, LRP achieves better detection performance over the state-of-the-art approaches tested.

On Learning Vehicle Detection in Satellite Video

“…Top, left: video frame of the SkySat-1 LasVegas video showing a city highway with multiple cars. Top, right: vehicle labelling provided by Zhang et al [34,33]. Bottom, left: the method's response (heat) map.…”

“…Satellite images capture large sceneries, usually dozens of square kilometers which introduce instead of a few visually large objects, thou-sands of tiny objects coming from hundreds of categories in a single image. At the same time these objects reduce in pixel size by orders of magnitude from 10 4 px to 10 2 px, to even 10 px for satellite video [34], depending on the camera's ground sample distance (GSD) 2 .…”

Vehicle Detection in Remote Sensing Image Based on Machine Vision

“…Gao et al [ 37 ] proposed a novel detection model, DE-CycleGAN, to enhance weak targets and achieve accurate remote sensing image detection. Zhang et al [ 38 ] proposed a three-step local proposal method (LRP) for the detection of live vehicles in satellite video. Shi et al [ 39 ] proposed a single-stage and anchorless detection method to detect oriented vehicles in high-resolution aerial images by linking coarse and fine feature maps output from different stages of the residual network through a feature pyramid fusion strategy.…”

Editorial for Special Issue “Remote Sensing for Target Object Detection and Identification”

“…Another group of papers [3][4][5][6][7][8][9] proposes object detection and recognition approaches that use images (or videos) acquired in the visible and near-infrared (VNIR) wavelength range, making use of the high (or very high) spatial resolution and high spectral content. Indeed, the latter are key features in order to identify shapes, thus enabling more reliable object detection and recognition.…”

“…Alternatively, Ma et al [4] employ CNNs to perform a stable and robust multi-model decision fusion, which jointly uses contextual features and object spatial structure information. Another interesting application of CNNs is described in Zhang et al [5], in which vehicle detection for traffic monitoring systems is performed using satellite video data. In contrast, Li et al [6] focus their work on the design of a parallel hardware architecture, based on multiple neural processing units (NPUs), for performing a power-efficient object detection by using CNNs.…”