Skip to main content
AI/ML Completed

Marine Object Detection

Deep learning model for detecting and classifying marine life and underwater objects using computer vision.

Timeline
Sep 2024 - Feb 2025
Category
AI/ML
Status
completed

Technologies Used

PyTorch YOLO Python OpenCV

Key Highlights

  • 95% detection accuracy
  • Real-time inference
  • Custom dataset of 10,000+ images

Overview

A custom-trained YOLO model for detecting and classifying marine objects in underwater imagery, designed to work with ROV camera feeds in real-time.

Architecture

The system uses a fine-tuned YOLOv8 model trained on a curated dataset of underwater images, with classes including fish species, coral types, marine debris, and infrastructure components.

Results

The model achieves 95% [email protected] on the test set using real-time inference at 30+ FPS on an NVIDIA Jetson platform.