Human Lens

A platform for businesses or groups training AI to crowdsource large pre-tagged image datasets using the blockchain

Comment

Inspiration

Some of our team members worked over the summer at small companies looking to get into the AI analytics space. We've noticed that acquiring the raw data required to train machine learning models is a significant barrier of entry into the industry.

What it does

Our platform connects business looking for large curated image data sets and freelance photographers from a web app to a an android app, then sets up a smart contract on the blockchain to trade images for financial compensation in the form of cryptocurrency . Organizations can use a web application to post queries detailing the type of image accepted, the number of images required, and the compensation per image. Once they've paid the freelancer they can access a .zip folder with their images. Freelancers can download an Android app and see a feed of queries from organizations, and upload images for specific queries. Once the images have been validated, they can connect their Coinbase wallet and receive compensation in the form of ether (cryptocurrency).

How I built it

We used Android Studio to design the Front End of the android app, and Angular.JS to design the front end of the web app. We used Firebase for our Backend, and Coinbase to connect Ethereum wallets with the rest of the app. Finally we used the Solidity language and IDE to write and test am Ethereum Smart Contract without uploading it to the blockchain.

Challenges I ran into

We ran into several issues integrated coinbase with our front and back end, and have issues writing to the Firebase backend from the Android app.

Accomplishments that I'm proud of

Our team had no prior experience in web development or blockchain, and have never used the Coinbase API. Nonetheless we have well designed web app, blockchain smart contract, and integration with Coinbase Wallet.

What I learned

We've learned a lot about how to build platforms integrating web and android applications, and automating file transfer in the back end. We've learned how to create and test ethereum smart contracts, and how to integrate and effectively use the blockchain.

What's next for Human Lens

We'd like to integrate a visual recognition into our backend to validate image submissions to improve the quality of the data sets our platform builds.

Share this project:

Updates