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

Object detection techniques that achieve state-of-the-art detection accuracyemploy convolutional neural networks, implemented to have optimal performancein graphics processing units. Some hardware systems, such as mobile robots,operate under constrained hardware situations, but still benefit from objectdetection capabilities. Multiple network models have been proposed, achievingcomparable accuracy with reduced architectures and leaner operations. Motivatedby the need to create an object detection system for a soccer team of mobilerobots, this work provides a comparative study of recent proposals of neuralnetworks targeted towards constrained hardware environments, in the specifictask of soccer ball detection. We train multiple open implementations ofMobileNetV2 and MobileNetV3 models with different underlying architectures, aswell as YOLOv3, TinyYOLOv3, YOLOv4 and TinyYOLOv4 in an annotated image dataset captured using a mobile robot. We then report their mean average precisionon a test data set and their inference times in videos of differentresolutions, under constrained and unconstrained hardware configurations.Results show that MobileNetV3 models have a good trade-off between mAP andinference time in constrained scenarios only, while MobileNetV2 with high widthmultipliers are appropriate for server-side inference. YOLO models in theirofficial implementations are not suitable for inference in CPUs.

Detecting soccer balls with reduced neural networks: a comparison of multiple architectures under constrained hardware scenarios