3M Hackathon: Machine learning for every developer

Hackathons give organizations an opportunity to focus engineering resources on solving challenges or driving bursts of innovation. On July 27, 2020, 3M hosted a week-long virtual 3MHackathon20: The Curiosity Code 2.0 where 800 3Mers from across the globe came together to solve real-world challenges using automation, machine learning, and various other technologies. In this hackathon, Amazon Web Services (AWS), the official partner, hosted a mini hack called the “AI/ML Buildout” for 6 teams and 70 participants over two days.

3M Hackathon history
The event was in its third year and has gotten bigger every year. The event grew out of an Alexa Challenge hosted by the 3M CRSL on campus in 2018, which brought more than 100 3Mers – again, many without software backgrounds – to explore how 3M can use interactive voice technology to better serve customers in the future. In 2019, more than 600 people took part across the globe, with in-person teams in Minnesota, as well as in the United Kingdom, Germany, China, and Japan. There were even a couple of customers attending for help on their projects. The format gives people dedicated time to focus on projects – and the access to experts to work through challenges that come up.

“The event resonated through the organization and internal partners we work with. So we decided to turn this into an annual event,” said David Frazee, technical director of 3M CRSL.

AWS has been an integral part of the hackathon and has been the event partner since 2018. “3M Hackathons are always very special to AWS team. Last year we did an AWS DeepRacer event where 150 3Mers got together at St. Paul RiverCentre hall to explore how 3M can use AWS services to better serve customers in the future. Due to the pandemic this year, we had to switch to Virtual Hackathon. Keeping the relevance, we designed the hackathon around use of Personal Protective Equipment (PPE) and AWS technology usage around them. The idea was to give the power of AWS services to users and help them scale quickly and one does not need to be an expert to learn it”, says Vivek Sethi Customer Solutions Manager at AWS.

The Hack
As part of the challenge, participants were asked to build a Face Mask Detection Platform to unlock a virtual Personal Protective Equipment (PPE) cabinet using AWS’s AI/ML services. Participants, assisted by the AWS facilitators, built a Face Mask Detection Platform using services like Amazon Rekognition, a service that makes it easy to add image and video analysis to applications that requires no machine learning expertise and Amazon SageMaker, a fully managed service that provides every developer and data scientist with the ability to build, train, and deploy machine learning (ML) models quickly. The challenge was to recognize if a user is wearing a mask and wearing it the right way. Subsequently, the participants unlocked the PPE cabinet with Amazon Rekognition and an object-detection system to recognize the products featured inside the cabinet using AWS Lambda, a service that lets you run code without thinking about servers. Finally, the participants built a system to translate and hear the content scanned from the cabinet’s inspection report document read back to them. This was built on Amazon Textract, a fully managed machine learning service that automatically extracts text and data from scanned documents, and Amazon Comprehend, a natural language processing (NLP) service that uses machine learning to find insights and relationships in text.

The winning criteria included time to completion and prediction accuracy of the trained model. The winning team completed the challenge in 45 minutes and used all of the AWS services recommended to train the model accurately to recognize if a user is wearing a mask and wearing it the right way. “Thank you for the team session today and I learned a lot! I hope to keep using these services with AWS in Future”, said one participant.

The AWS Prototyping team hosted an “Ask the Expert” chat room featuring solutions architects as facilitators to answer general queries about AWS and its services. The hack was a collaborative learning experience, letting every day users learn about AWS services and how easy it is to spin up instances. A builder or developer does not need to be an expert in machine learning in order to build something new with AWS services.

“Despite the headwinds we are all facing in 2020, the “3MHackathon20: The Curiosity Code 2.0,” was a success. Building on last year’s AWS DeepRacer competition, this year AWS played an even bigger, more integrated part of the hack. The AWS team setup deep dive hacks around several pressing topics and ideas across the globe. They worked with us side-by-side to master new ML services in materials development, air flow predictions, automated manufacturing, and voice-powered research and testing, just to name a few. If one of our goals was to show how to create value at the speed of AWS, then we just nailed it – together,” said Frazee.
If you missed the event, AWS offers pre-trained AI Services with ready-made intelligence for applications and workflows to help improve business outcomes – based on the same technology used to power Amazon’s own businesses. No prior machine learning skills is required. Examples include advanced text analytics, automated code reviews, chatbots, demand forecasting, document analysis, text to speech, transcription, and real-time translation. Find out more here – https://aws.amazon.com/machine-learning/