2021 Microsoft Azure Machine Learning ROV Challenge
The MATE ROV Competition is pleased to announce the 2021 Microsoft Azure Machine Learning ROV Challenge!
Your mission: Harness the power of cloud computing to identify and quantify organisms in benthic video transects. Start with pre-built Artificial Intelligence models like Microsoft Computer Vision and Video Indexer OR take it to the next level by building your own machine learning models.
The MATE ROV Competition and Microsoft are challenging EXPLORER and RANGER class teams to use the Microsoft Azure AI Platform to create a program that is able to evaluate organisms in a video transect. Annotated benthic video transects will be provided to teams for practice purposes. For the challenge, teams will be required to create an AI program that takes in a video transect and outputs data based on a specific task. Specific tasks may include having the program:
- Quantify the numbers of a species seen throughout the entire video
- Quantify the numbers of different species seen at one moment
- Determine the size of an organism
- Highlight a species every time it is encountered in the video
- Determine the number of times a designated species is encountered in the video
- Differentiate between an organism and debris/trash
This is in the spirit of an open hackathon. Teams will be provided several video transects, each with an assigned task. Pick your preferred platform, libraries, & developer tools – it’s all up to you with the distributed computing scale of Azure. Teams that participate will be judged on the accuracy of their counts or analysis.
Teams should consider attempting all of the specific tasks suggested in the above bullet points, and any others they deem plausible. Additional video transects will be linked to this page for additional practice.
Participation in Microsoft Azure Challenge is OPTIONAL, but it is intended to foster creativity, develop skills for “the future of work,” and provide a new, fun approach to competition tasks. It is a way for teams to learn and grow, rather than compete for maximum points.
Each participating team member may register for Azure for Students to get $100 cloud credit plus free developer tools like Visual Studio Code and Python libraries for data science. Teams that qualify to attend the MATE ROV World Championship will receive additional Azure cloud credit with the value to be determined at the time of the event based on the number of teams selected.
Are there prizes for this challenge?
In addition to learning / applying your career-building skills in Cloud Computing – the #1 in-demand “hard” skill according to LinkedIN – the MATE ROV Competition will provide prizes for the top teams that compete in the Microsoft Azure challenge.
Timeline and Scoring
The opening round of the Microsoft Azure challenge will occur in spring 2021. All competing teams will be provided with a video transect, a task, and a deadline. Teams will be evaluated on the accuracy of their program to complete the given task. The top ten teams from EXPLORER, and the top ten teams from RANGER will advance from the initial round to the semi-final round. The top four teams from EXPLORER and the top four teams from RANGER will advance from the semi-final round to the finals. The final round will take place during the World Championship, with the format being tournament style brackets. Teams will be provided with a new video transect and/or specific task each round.
RANGER and EXPLORER teams that do not qualify for the World Championship may still compete in the Microsoft Azure challenge. Teams not at the World Championship event can showcase their programming through Zoom, Microsoft teams, or other video conferencing platforms.
How do you get started?
Submit your intent to participate to the 2021 Microsoft Azure Machine Learning ROV Challenge by March 1st, 2021.
How do you access Microsoft Azure?
Start with Azure for Students, which provides $100 cloud credit for students that verify with their accredited school’s email address. Enrolling does not require a credit card. There is no obligation to pay for Azure in the future. If you are unable to verify online because your school is not supported, contact MATE for a one-time code.
Each student in the project team may obtain Azure for Students. For example, if you have a team of 7 students, this team can collectively receive $700 worth of credit but each $100 increment cannot be combined into one Azure account. But one student can give access to other team members through role-based access control so practically this means that students can experiment independently and then contribute to one designated student’s account (representing the team’s contribution).
Teams are not required to use the Azure for Students offer to participate. Your team can choose to pay for additional Azure credit. For example, if your team has been sponsored by other organization or group, they may choose to support you by paying for Azure independently.
What technical support and resources are available?
New to cloud development? Start with Microsoft Learn for Students which offers helpful tutorials in data science and cloud app development. Looking for something specific? Go to the complete Azure technical docs. Or try these specifically:
- AI & Machine Learning services (pre-built AI and flexible tools for building models)
- Data Science Virtual Machine (pre-built VM for Linux or Windows)
- Python SDKs and libraries (for your preferred IDE)
- Using Python in Visual Studio Code (cross-platform IDE with extensions)
- Azure Machine Learning (includes low/no-cost ML Designer and Jupyter Notebooks with Python runtime)
- SQL Database as a service (for relational database, non-relational options available too)
- Power BI as a service with Azure (for data visualization)
Limited technical support will be available prior to the MATE ROV Finals event. Please consult the documents above. However, if you are stuck and need some help, post your questions on the Azure Challenge forum located here: https://forums.marinetech2.org/viewforum.php?f=27.
HINT: Check out Coral Morphologic: