Log in to save to my catalogue

A Semi-Supervised Domain Adaptation Method for Sim2Real Object Detection in Autonomous Mining Trucks

A Semi-Supervised Domain Adaptation Method for Sim2Real Object Detection in Autonomous Mining Trucks

https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_doaj_primary_oai_doaj_org_article_783d3013c32a4117ae8befed0723db0e

A Semi-Supervised Domain Adaptation Method for Sim2Real Object Detection in Autonomous Mining Trucks

About this item

Full title

A Semi-Supervised Domain Adaptation Method for Sim2Real Object Detection in Autonomous Mining Trucks

Publisher

Switzerland: MDPI AG

Journal title

Sensors (Basel, Switzerland), 2025-02, Vol.25 (5), p.1425

Language

English

Formats

Publication information

Publisher

Switzerland: MDPI AG

More information

Scope and Contents

Contents

In open-pit mining, autonomous trucks are essential for enhancing both safety and productivity. Object detection technology is critical to their smooth and secure operation, but training these models requires large amounts of high-quality annotated data representing various conditions. It is expensive and time-consuming to collect these data during...

Alternative Titles

Full title

A Semi-Supervised Domain Adaptation Method for Sim2Real Object Detection in Autonomous Mining Trucks

Identifiers

Primary Identifiers

Record Identifier

TN_cdi_doaj_primary_oai_doaj_org_article_783d3013c32a4117ae8befed0723db0e

Permalink

https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_doaj_primary_oai_doaj_org_article_783d3013c32a4117ae8befed0723db0e

Other Identifiers

ISSN

1424-8220

E-ISSN

1424-8220

DOI

10.3390/s25051425

How to access this item