https://ar.iub.edu.bd/handle/11348/1043 Sat, 27 Jun 2026 09:45:19 GMT 2026-06-27T09:45:19Z https://ar.iub.edu.bd/handle/11348/1186 Machine-Vision-Based Bangladeshi Geographical Indication (GI) Dessert Recognition Bhuiyan, Shakibur Rahman; Kundu, Dip; Siam, Md Tamzed Hossain Geographical Indication (GI) certification preserves products whose identity and value are linked to their geographical origin. Despite the cultural and economic importance of Bangladeshi GI-certified desserts, no benchmark dataset or automated recognition system previously existed, creating risks of counterfeiting and mislabeling. This thesis addresses the gap by introducing BD-GI-Desserts13, a dataset of 11,692 images across 13 GI dessert classes, and proposing a machine-vision framework for fine-grained dessert recognition. The study develops a lightweight Custom Convolutional Neural Network (CCNN) and compares it with transfer-learning and traditional machine learning models. Experimental results show that the CCNN achieves 99.37% test accuracy while remaining computationally efficient for real-time web and mobile applications. The proposed framework provides a reproducible baseline for future GI food recognition research in Bangladesh and serves as a decision-support tool rather than a legal certification system. Sat, 25 Apr 2026 00:00:00 GMT https://ar.iub.edu.bd/handle/11348/1186 2026-04-25T00:00:00Z https://ar.iub.edu.bd/handle/11348/1044 Dual-Task Real-Time Low-Light Lane and Pothole Detection for Resource-Constrained Environments Md Iftekharul, Alam Lane detection and road hazard awareness are crucial for ensuring safety in autonomous driving and Advanced Driver-Assistance Systems (ADAS). These systems rely heavily on clear visual cues, which are often compromised in low- light driving scenarios. The challenge is especially pronounced in low- and middle-income countries (LMICs), where poorly illuminated roads, faded lane markings, and unmaintained sur- faces frequently co-occur. Under such conditions, conventional single-model detectors trained for daytime environments degrade sharply, as lane cues and pothole textures often compete in the same field of view. To address this, we present a lightweight dual- model pipeline that integrates a low-light enhancement front end with an OpenCV-based lane delineation pipeline and a YOLOv12 detector for pothole localization. The models run in parallel on shared inputs, and their outputs are fused to generate a unified lane geometry and hazard map in a single pass. The architecture is optimized for modest compute and memory budgets, enabling deployment in resource-constrained settings while maintaining high throughput. Evaluated on evening-time urban road scenes from Bangladesh, achieves 88.7potholes and 89.3FPS on NVIDIA GTX 1050Ti, outperforming a single-detector baseline. These results highlight the potential of our approach for practical, real-time ADAS perception in underserved regions. Index Terms—Low-light imaging, Lane detection, Pothole de- tection, YOLOv12, OpenCV, Image enhancement, Edge comput- ing, Autonomous driving Sun, 04 Jan 2026 00:00:00 GMT https://ar.iub.edu.bd/handle/11348/1044 2026-01-04T00:00:00Z