Transfer learning improves VLC indoor localization on Bosch data
Original title: A Transfer Learning-Driven Methodology for Efficient and Sustainable Indoor Localization Using VLC
The authors report 47 percent better accuracy than a non-transfer model and model efficiency up to 67.5 percent in high noise.
Researchers propose transfer learning for visible-light indoor localization so sites can avoid repeated data collection and heavy calibration when lighting and obstacles change. They introduce a model-efficiency metric combining accuracy, energy, data use, and transfer gain. On a real-world dataset from a Bosch factory, they report 47 percent better accuracy than a conventional model and model efficiency up to 67.5 percent versus 10.4 percent under high noise.
- The method targets lighting fluctuations and physical obstacles.
- Transfer-learning models reached 67.5 percent model efficiency versus 10.4 percent.
- Evaluation uses a dataset collected at a Bosch factory.
- The metric also scores energy use, data efficiency, and transfer gain.
From the original article
1 min read at arxiv.orgThe rapid expansion of the Internet of Things enables seamless collaboration among connected devices, making indoor localization a critical component of industrial automation. However, conventional localization techniques struggle in dynamic environments.
By Masood Jan · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.