University Building Recognition Dataset in Thailand for the mission-oriented IoT sensor system

2025-12-05

AI Summary

To address the lack of localized building recognition datasets and decentralized training frameworks in edge IoT environments, this paper introduces CUBR—the first building recognition dataset tailored to the Chulalongkorn University campus in Thailand—and proposes WAFL-ViT, an end-to-end framework integrating Wireless Ad-hoc Federated Learning (WAFL) with Vision Transformers (ViT). Leveraging device-to-device communication and edge computing, WAFL-ViT enables collaborative model training across heterogeneous edge devices without relying on a central server. Experimental results demonstrate that WAFL-ViT achieves an 8.3% absolute accuracy improvement over local isolated training on CUBR, validating both the dataset’s practical utility and WAFL’s effectiveness for task-oriented edge intelligence. This work contributes a publicly reusable, region-specific visual perception dataset and a lightweight, communication-efficient distributed learning paradigm, advancing deployable vision systems for resource-constrained edge IoT deployments.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Adversarial Attacks & Robustness
• Intelligent Robots: Cognitive Robotics

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Many industrial sectors have been using of machine learning at inference mode on edge devices. Future directions show that training on edge devices is promising due to improvements in semiconductor performance. Wireless Ad Hoc Federated Learning (WAFL) has been proposed as a promising approach for collaborative learning with device-to-device communication among edges. In particular, WAFL with Vision Transformer (WAFL-ViT) has been tested on image recognition tasks with the UTokyo Building Recognition Dataset (UTBR). Since WAFL-ViT is a mission-oriented sensor system, it is essential to construct specific datasets by each mission. In our work, we have developed the Chulalongkorn University Building Recognition Dataset (CUBR), which is specialized for Chulalongkorn University as a case study in Thailand. Additionally, our results also demonstrate that training on WAFL scenarios achieves better accuracy than self-training scenarios. Dataset is available in https://github.com/jo2lxq/wafl/.

Problem

• Develops a specialized building recognition dataset for Thai university IoT systems.
• Tests Vision Transformer in wireless ad hoc federated learning for edge devices.
• Compares collaborative WAFL training accuracy against individual self-training methods.

Innovation

• WAFL-ViT uses Vision Transformer for edge training
• Device-to-device communication enables collaborative federated learning
• Mission-specific dataset CUBR improves building recognition accuracy