Airtable Rekognition

Airtable rekognition allows you to run a multitude of deep learning capabilities over the images in your airtable base. Detect objects, recognise celebrities and detect explicit content.

Comment 1

Inspiration

I loved the idea of running deep learning over images in my airtable base, automating what would be a very human centric task.

What it does

Airtable-rekognition uses the AWS rekognition API to detect objects in your image attachments such as cars, people and more. The block will also recognise celebrities and NSFW content, all of which will be displayed in a dialog showing the results of your deep learning report.

How I built it

I built it using the nodeJS AWS SDK as well as some HTTP libraries and of course the wonderful blocks API!

Challenges I ran into

Streaming the bytes of the images using fetch so they could be fed into the rekognition API.

Accomplishments that I'm proud of

Providing powerful functionality to a platform I love in airtable. I was happy to provide a simple way to combine 2 powerful tools

What I learned

  • How to use the blocks SDK
  • Streaming bytes directly to AWS rekognition
  • Some cool ideas to implement when writing react hooks in future. Airtable has implemented a fantastic, declarative API that I read some of the source code of. It helped me with some ideas for designing hooks in future.

What's next for Airtable Rekognition

  • global settings to enable which image analysis reports you want to see
  • Write the results back into another airtable table
Share this project:

Updates