This GitHub repository showcases our team's work for the MATE / NOAA Ocean Exploration Video Challenge. It includes relevant files, documentation, and resources related to our computer vision team's participation in the competition.
The MATE / NOAA Ocean Exploration Video Challenge is a global competition that aims to promote innovative technologies and solutions for analyzing underwater videos. The challenge involves analyzing underwater video footage to identify and classify marine life species accurately.
-
Nira_explanation_2023.pdf: This document provides a comprehensive overview of our computer vision team's approach, methodologies, and results achieved during the challenge.
-
Nira_spreadsheet_2023.xlsx: We have included a spreadsheet containing the resulting bounding box coordinates for each of the analyzed videos. These coordinates represent the detected marine life species and their locations in the videos.
-
2023-OER-MATE-ROV-Computer-Coding-Challenge_FINAL.pdf: The official problem statement for the MATE / NOAA Ocean Exploration Video Challenge is available here for reference.
-
Results PDF File: This document showcases detailed analysis and insights from our computer vision algorithms and the overall performance of our model.
-
YOLOv7 Results Folder: We have modified the official YOLOv7 repository's
detect.pyfile to tailor it to our specific needs during the competition. The folder contains the modifieddetect.pyfile along with relevant graphs and analysis derived from the YOLOv7 model. -
Winners - 2023 Awards-MATE ROV Competition.v2.pdf: The results published for the 2023-MATE-NOAA-Ocean-Exploration-Computer-Coding-Challenge
Check out the resulting videos generated by our computer vision algorithms on our YouTube playlist: Link to Youtube playlist
The final model used for the MATE / NOAA Ocean Exploration Video Challenge is hosted on Google Drive. You can access it here: Link to Google Drive
Detailed instructions for running the model and analyzing the underwater videos are provided in the "Nira_explanation_2023.pdf" file (Documentation PDF File). Follow the step-by-step guide to replicate our results and analyze your own underwater video files or real-time video feeds.
Our computer vision team, comprising of myself, R Soorya Narayanan and Abhishek M J (github), dedicated ourselves to this challenge, as part of Team Nira from The AUV Society, IIITDM Kancheepuram, and our combined efforts and commitment have enabled us to secure the 3rd position worldwide in this prestigious competition.
We extend our deepest gratitude to all those who have supported us on this remarkable journey. Special thanks to our mentors and advisors for their invaluable guidance and encouragement throughout the competition.