KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Search Results

Now showing 1 - 6 of 6

Some of the metrics are blocked by your
consent settings
YOLO Based IoT Tracking for Academic Labs on Raspberry Pi
(2026-06-16)
Jiamrachada, Tamakorn
;
Nonsiri, Sarayut
;
Kamin, Pichitchai
;
Nanthajirapong, Nathaphon
Academic IoT laboratories often rely on shared equipment, making efficient borrow-return management essential. Conventional management methods depend on manual recording, which can cause verification delays, data entry errors, and increased staff workload. This study proposes a YOLO based IoT equipment tracking system that uses a camera to detect and count devices inside student equipment boxes for borrow-return recording and inventory monitoring. The system runs on a Raspberry Pi 5 for real-time edge-based processing, while detection results are stored in a database and displayed through a web-based dashboard. Experimental results show that the YOLO12n model achieved an F1-score of 0.996 with a real-time inference speed of 12.01 FPS, demonstrating the system's effectiveness in reducing human error and improving laboratory inventory management efficiency.
Some of the metrics are blocked by your
consent settings
Library Seat Hogging Detection Using Hybrid Real and AI-Generated Data
(2026-01-01)
Viwatanawatanakarn, Natchanon
;
Cooharojananone, Nagul
;
Muangsin, Veera
;
Tea-Makorn, Pin Pin
;
Atchariyachanvanich, Kanokwan
Efficient management of library seating resources is a critical challenge in educational institutions, often hindered by 'seat hogging' behaviors where users occupy spaces with personal belongings without actual occupancy. Traditional manual inspections are labor-intensive and inefficient. This paper proposes an automated seat occupancy detection system utilizing existing CCTV infrastructure and Computer Vision techniques. We employ YOLOv8, a state-of-the-art object detection model, to identify two key classes: persons and belongings. To address the challenge of limited real-world datasets for specific library environments, we introduce a data augmentation strategy using AI-generated synthetic data produced by a generative model (Gemini 2.5 Pro). A rule-based algorithm is integrated to analyze the spatiotemporal relationship between detected persons and belongings, enabling the system to distinguish between 'occupied,' 'vacant,' and 'hogged' states effectively. Experimental results demonstrate that the proposed hybrid dataset approach enhances detection performance, providing a scalable and cost-effective solution for smart library management. Furthermore, a pilot system evaluation yielded an overall accuracy of 91.62%, validating the system's effectiveness for real-world deployment.
Some of the metrics are blocked by your
consent settings
Unified Multimodal-Multitask Learning for Vehicle Damage Assessment in Insurance Applications
(2026-01-01)
Phuengpanyaloet, Wongsapat
;
Pasupa, Kitsuchart
;
Angsarawanee, Thanatwit
;
Chetprayoon, Panumate
;
Sakdejayont, Theerat
Automated vehicle damage assessment requires both precise localization and clear textual reporting. While existing methods typically treat these as separate tasks, the trade-offs of unified multimodal-multitask learning in this domain remain underexplored. This paper conducts a comparative study between a unified vision-language framework, Generative Region-to-Text Transformer (GRiT), and single-task baselines derived from GRiT by isolating the detection and captioning components. We adapt GRiT to the insurance domain using a dataset enriched with vehicle part annotations and structured damage descriptions. Experimental results demonstrate that the unified model achieves competitive detection performance (F<inf>1</inf>-score: 0.54), slightly outperforming the detection baseline model. Crucially, it significantly surpasses the caption baseline model in description quality (METEOR: 0.75, ROUGE: 0.70, BLEU: 0.46), confirming that object-level visual grounding is essential for accurate reporting. These findings indicate that unified multimodal learning enhances semantic interpretation without compromising localization accuracy, offering a promising direction for automated insurance workflows.
Some of the metrics are blocked by your
consent settings
Detection of OCR Misclassifications in ALPR via Rapid Prototyping
(2025-01-01)
Oo, Saung Hnin Pwint
;
Yoshikawa, Yuki
;
Kurimoto, Ikusaburo
This paper originates from the limitations of traditional Optical Character Recognition (OCR) systems, particularly in recognizing multilingual texts (English and Thai characters) within Automatic License Plate Recognition (ALPR). A common challenge in current OCR models is character confusion, such as misreading between “0” and “O”, and “8” and “B,” which can significantly impact character recognition accuracy. Since OCR performance directly affects the overall effectiveness of ALPR systems, such misclassifications lead to incorrect vehicle identification. Therefore, this paper introduces a rapid prototyping approach for ALPR, focusing on multilingual license plates, especially Thai license plates that include both Latin (English) and non-Latin (Thai) characters. It aims to emphasize the identification and analysis of OCR misclassifications in how characters are recognized by OCR.
Some of the metrics are blocked by your
consent settings
Train Collision Avoidance System by X-Band FMCW Radar and Long Range Camera Image Processing
(2024-01-01)
Pongthavornkamol, Tiwat
;
Klangnurak, Thanirat
;
Manasummakij, Prateep
;
Wisadsud, Sodsai
;
Manatrainon, Supatra
Sensor fusion between X-Band radar and long-range Pan-Tilt-Zoom (PTZ) camera for train collision avoidance system is proposed. The method of Constant False Alarm Rate (CF AR) is applied for the radar signal processing to improve the correct result of radar detection. The edge detection and object detection by digital image processing are applied to long range camera system for classification of identity and position of the object. Experiments of obstacle detection by radar and camera are performed. The proposed work will benefit the safety improvement of locomotive transportation application in which its details are presented and discussed in this paper.
Some of the metrics are blocked by your
consent settings
Weapon Detection in X-ray Image of Baggages
(2024-01-01)
Kundilokovit, Piyapat
;
Thaweechoklertchaikul, Rimthaweep
;
Anuntachai, Anuntapat
Due to the daily commutes of people by MRT trains, following the shooting incident at Paragon, the MRT system has implemented bag searches before entering the stations to look for concealed or hidden weapons. These searches are conducted manually, which sometimes may not be thorough enough and can take a significant amount of time. Especially during peak hours when many people are using the MRT, it is possible for some individuals to pass through the station without being searched. Such actions can render the security measures ineffective. Therefore, this paper proposes a study to find ways to address these issues. From the study and comparison of object detection processes for risky items, such as sharp objects or guns, in X-ray images of luggage, it was found that models such as CNN, RCNN, Detectron, RetinaNet, and Yolo achieved excellent results in object detection and recognition. The organizers plan to apply object detection techniques and improve the existing methods for detecting objects in X-ray images to be more efficient and accurate, capable of identifying a variety of risky items.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results