English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Various autonomous or assisted driving strategies have been facilitatedthrough the accurate and reliable perception of the environment around avehicle. Among the commonly used sensors, radar has usually been considered asa robust and cost-effective solution even in adverse driving scenarios, e.g.,weak/strong lighting or bad weather. Instead of considering to fuse theunreliable information from all available sensors, perception from pure radardata becomes a valuable alternative that is worth exploring. In this paper, wepropose a deep radar object detection network, named RODNet, which iscross-supervised by a camera-radar fused algorithm without laborious annotationefforts, to effectively detect objects from the radio frequency (RF) images inreal-time. First, the raw signals captured by millimeter-wave radars aretransformed to RF images in range-azimuth coordinates. Second, our proposedRODNet takes a sequence of RF images as the input to predict the likelihood ofobjects in the radar field of view (FoV). Two customized modules are also addedto handle multi-chirp information and object relative motion. Instead of usinghuman-labeled ground truth for training, the proposed RODNet iscross-supervised by a novel 3D localization of detected objects using acamera-radar fusion (CRF) strategy in the training stage. Finally, we propose amethod to evaluate the object detection performance of the RODNet. Due to noexisting public dataset available for our task, we create a new dataset, namedCRUW, which contains synchronized RGB and RF image sequences in various drivingscenarios. With intensive experiments, our proposed cross-supervised RODNetachieves 86% average precision and 88% average recall of object detectionperformance, which shows the robustness to noisy scenarios in various drivingconditions.

RODNet: A Real-Time Radar Object Detection Network Cross-Supervised by Camera-Radar Fused Object 3D Localization