THESIS

Intergration of Clache-Base Image Enhancement Techniques and YOLOv3 Model on Edge AI System

for Real-Time Underwater Fish Detection

ABSTRACT

Underwater fish detection is essential for marine ecosystem monitoring, yei detection performance is often degraded by low contrast, bluish-green color dominance, turbidity, and non-uniform illumination. This paper proposes an end-to- end underwater fish detection pipeline that integrates CLAHE-based image enhancement with YOLOv3 object detection and validates its deployment on an NVIDIA Jetson Orin Nano edge-AI device using a Streamlit interface. Four CLAHE- based enhancement ariants are evaluated: CLAHE, CLAHE with Unsharp Masking (USM), CLAHE with High-Frequency Emphasis Filtering (HEF), and CLAHE combined with Percentile Blending. Image quality is assessed on a 10% subset of the DeepFish dataset using LOE, UIOM, and UCIOE metrics. The results show that CLAHE with Percentile Blending provides the best trade-off, achieving the lowest LOE (212.021) and the highest UIOM (3.949) and UCIOE (0.508) Detection performance is evaluated on two domains: the DeepFish dataset and a locally collected dataset from Sabang, Aceh (Indonesia) to assess domain shift. Or DeepFish, image enhancement yields an incremental improvement from 96.15% ta 97.05% mAP@0.5, accompanied by a reduction in false negatives (198 to 137) and an increase in mean IoU (74.39% to 75.09%). In contrast, direct testing on the loca dataset without adaptation results in a severe performance drop (6.07% mAP@0.5) After fine-tuning on the local dataset using a 70/15/15 train/validation/test split performance improves ubstantially to 92.64% mAP@0.5 on original inputs ana further to 93,74% on enhanced inputs, with a recall of 0.92. The proposed system operates fully offline on the Jetson Orin Nano and provides detection outputs via the Streamlit interface.

PUBLICATION

• Comparative Analysis of Liver Disease Classification Using Ensemble Learning and Feature Selection (link)
• Utilizing Edge Device and Streamlit for Comparative Evaluation of CLAHE-Based Underwater Image Enhancement Techniques (link)
• Optimizing Pomegranate Growth Stage Detection Using YOLOv8 and Hybrid Machine Learning Approaches (link)