Fishing_Event_Detection_Final_Presentation.pptx

Vmjfcgjmn

1. Fishing Event Detection and Species Classification Using Computer Vision and Artificial Intelligence for Electronic Monitoring Presented by: [Your Name] Institution: [Your Institution] Guide: [Guide’s Name]

2. Introduction • Fisheries require accurate catch monitoring for sustainability. • Manual methods are expensive and inconsistent. • AI-based automated detection can improve monitoring efficiency.

3. Problem Statement • Human observers are costly and limited. • Reviewing EM videos manually is time-consuming. • Need for automated AI-based fish detection and classification.

4. Research Objectives • Develop an AI-based system for fish detection. • Implement deep learning models for species classification. • Improve efficiency in fisheries management.

5. Literature Review • Past research explored AI for fish detection. • Methods included R-CNN, YOLO, Mask R-CNN. • Our study improves accuracy with YOLOX and ConvNeXt models.

6. Methodology 1. **Fish Detection:** YOLOX deep learning model. 2. **Tracking:** Kalman filter & Hungarian algorithm. 3. **Classification:** ConvNeXt model for species identification. 4. **Evaluation:** Precision, recall, and accuracy metrics.

7. System Architecture • Fish detection using deep learning. • Tracking of fish movements. • Classification into different species. • Outputs: Count, species, and event analysis.

8. Implementation • Tools: Python, TensorFlow, PyTorch. • Datasets: EM videos from commercial fishing vessels. • Training: YOLOX for detection, ConvNeXt for classification. • Optimization: Data augmentation and cross-validation.

9. Results & Analysis • Fish detection accuracy: 87%. • Event detection precision: 81%. • Fish species classification accuracy: 91.11%. • Effective for sustainable fisheries management.

10. Challenges & Limitations • Poor video quality due to weather conditions. • Data imbalance in fish species dataset. • Tracking issues with overlapping fish. • Need for better real-time processing.

11. Conclusion & Future Scope • AI-based fish detection improves monitoring efficiency. • Automates fisheries management processes. • Future work: Expand dataset, improve real-time tracking, and apply to more vessels.

12. References & Acknowledgments • Research paper: Fishing Event Detection using AI. • Authors: Muhammad Saqib et al. • Supported by CSIRO & Australian Fisheries Management Authority. • Thanks to the fishing vessel crew and reviewers.