ocean_shrimp

YouTube/iOS live stream coral reef (shrimp) to AWS Rekognition for label detection and serve results to django server

Comment

Team Name -> Diversity

Table Number -> 28

Team Names -> Tony Li, Charles Bowen Shi, Carter Mutian Gao, Ryan Gontarek

Inspiration

One of our close friends studies oceanography and coral reefs and we've always joked about how we would automate their job once amazon comes out with a submarine drone and an sdk for image recognition.

What it does

Takes a stream of frames from youtube live stream and sends them into AWS Rekognition for feature/label detection.

How we built it

Integrating youtube / iOS -> s3 -> Rekognition -> DynamoDB -> Django

Challenges we ran into

Automation takes time and Rekognition is so new a service there are not tutorials on how to set up a video stream with their SQS service. Even within the AWS documentation they suggest running a standard queue service for the live stream when in reality you need a FIFO queue to preserve order.

Accomplishments that we're proud of

It's our first time using a CI/CD development environment in AWS!

What we learned

How to stream live video from youtube to AWS. How to use s3, dyanmoDB, CI/CD in AWS.

What's next for ocean_shrimp

We will repurpose the tool for other coral reefs and create better fish detection models.

Built With

Share this project:

Updates