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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Modern object detectors usually suffer from low accuracy issues, asforegrounds always drown in tons of backgrounds and become hard examples duringtraining. Compared with those proposal-based ones, real-time detectors are infar more serious trouble since they renounce the use of region-proposing stagewhich is used to filter a majority of backgrounds for achieving real-timerates. Though foregrounds as hard examples are in urgent need of being minedfrom tons of backgrounds, a considerable number of state-of-the-art real-timedetectors, like YOLO series, have yet to profit from existing hard examplemining methods, as using these methods need detectors fit series ofprerequisites. In this paper, we propose a general hard example mining methodnamed Loss Rank Mining (LRM) to fill the gap. LRM is a general method forreal-time detectors, as it utilizes the final feature map which exists in allreal-time detectors to mine hard examples. By using LRM, some elementsrepresenting easy examples in final feature map are filtered and detectors areforced to concentrate on hard examples during training. Extensive experimentsvalidate the effectiveness of our method. With our method, the improvements ofYOLOv2 detector on auto-driving related dataset KITTI and more general datasetPASCAL VOC are over 5% and 2% mAP, respectively. In addition, LRM is the firsthard example mining strategy which could fit YOLOv2 perfectly and make itbetter applied in series of real scenarios where both real-time rates andaccurate detection are strongly demanded.

Loss Rank Mining: A General Hard Example Mining Method for Real-time Detectors