Conference Monitoring System based on Image Recognition in Rust

Industry

Technology

About Conf-Count

Conf-Count is a Conference Monitoring System in Rust that uses AWS services like S3 and Rekognition to get the similarity score of two images sent through the Rust application. The overall goal of this project is to monitor the conference attendees and keep track of all the conferences held in an organisation.

This project is based on Image Recognition, where all the registered user’s image is compared with the present user’s image using AWS Rekognition Service If any registered user is unable to attend the conference, so his conference status will remain Absent and an Email is sent to that user along with his Status and Talk Content.

Challenges

The major problems which we faced were as follows:

Solution

Interaction with AWS Rekognition was quite difficult through Rust Programming because it throws an error related to Runtime, so we provide Tokio runtime (tokio::runtime::Runtime) and spawn that Rekognition thread into (futures::sync::oneshot::spawn) method to work with CompareFaces API of Rekognition service.

Triggering RPi camera asynchronously was not the right decision, because asynchronous lets the program works with the previously captured image because it’s an asynchronous task and a new image is still in process. So we decided to make this triggering camera synchronous so that we can get the current image always. Handling Multipart was not that easy as we think, so we decided to club the (futures::stream) and (actix-web::multipart) to handle the multipart request.