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/

Zendy Open

Presents the pdf analysis as premium feature

PDF Analysis

Insights

Presents the summarisation as premium feature

Summarisation

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Correctly identifying crosswalks is an essential task for the drivingactivity and mobility autonomy. Many crosswalk classification, detection andlocalization systems have been proposed in the literature over the years. Thesesystems use different perspectives to tackle the crosswalk classificationproblem: satellite imagery, cockpit view (from the top of a car or behind thewindshield), and pedestrian perspective. Most of the works in the literatureare designed and evaluated using small and local datasets, i.e. datasets thatpresent low diversity. Scaling to large datasets imposes a challenge for theannotation procedure. Moreover, there is still need for cross-databaseexperiments in the literature because it is usually hard to collect the data inthe same place and conditions of the final application. In this paper, wepresent a crosswalk classification system based on deep learning. For that,crowdsourcing platforms, such as OpenStreetMap and Google Street View, areexploited to enable automatic training via automatic acquisition and annotationof a large-scale database. Additionally, this work proposes a comparison studyof models trained using fully-automatic data acquisition and annotation againstmodels that were partially annotated. Cross-database experiments were alsoincluded in the experimentation to show that the proposed methods enable usewith real world applications. Our results show that the model trained on thefully-automatic database achieved high overall accuracy (94.12%), and that astatistically significant improvement (to 96.30%) can be achieved by manuallyannotating a specific part of the database. Finally, the results of thecross-database experiments show that both models are robust to the manyvariations of image and scenarios, presenting a consistent behavior.

Automatic Large-Scale Data Acquisition via Crowdsourcing for Crosswalk Classification: A Deep Learning Approach