PixGuard

Smart image screening to protect platforms from unsafe content.

Comment

Inspiration

This project was inspired by the growing need to moderate user-uploaded images quickly and effectively. Manual review is time-consuming and not scalable, especially for small platforms. By using instant photo classification, this app helps ensure content safety while delivering a clean, efficient user experience, making digital spaces smarter, safer, and easier to manage.

What it does

This project provides a serverless photo recognition service that allows users to upload images, which are then analyzed using Amazon Rekognition. The service categorizes images as suitable or unsuitable based on predefined criteria.

How we built it

We used several AWS serverless services to stitch the product together:

  • Amazon Rekognition: For image analysis and recognition.
  • AWS Lambda: For serverless compute to process images.
  • Amazon S3: For storing images.
  • AWS Step Functions: For orchestrating the workflow of image processing.
  • AWS CDK: For defining the cloud infrastructure as code.
  • AWS IAM: For managing permissions and roles for the Lambda function and Step Functions.
  • AWS CloudFormation: For deploying the infrastructure defined in the CDK.

Challenges we ran into

  • Setting up the AWS CDK environment and ensuring all dependencies were correctly installed.
  • Configuring the necessary IAM roles and permissions for the Lambda function and Step Functions.
  • Handling large image uploads and ensuring they are processed efficiently.
  • Implementing error handling and retries for the image processing workflow.

Accomplishments that we're proud of

  • Successfully implemented a serverless architecture using AWS Lambda and Step Functions.
  • Leveraged Amazon Rekognition for advanced image analysis and moderation.
  • Automated deployment using CDK
  • Built a user-friendly frontend client using React for seamless image uploads and result display.

What we learned

In this project, we learned how to use AWS services like Lambda, Step Functions, and Rekognition to build a serverless photo recognition application. We also gained experience with the AWS CDK for infrastructure as code.

What's next for PixGuard

  • Integrate a feedback loop to improve classification accuracy over time.
  • Enable batch upload and processing of images.
  • Extend support to video moderation using Amazon Rekognition Video.

Built With

Share this project:

Updates