Building an Easter Egg Detector with AWS Free Tier
The author built a Streamlit app that uses Amazon Rekognition's standard object detection and custom model training to detect Easter eggs in images, all within the AWS Free Tier.
Why it matters
This project demonstrates how developers can leverage the AWS Free Tier to experiment with and build AI/ML-powered applications, even for niche use cases like Easter egg detection.
Key Points
- 1Leveraged AWS Free Tier to experiment with Amazon Rekognition's computer vision capabilities
- 2Developed a Streamlit app with two modes: standard object detection and custom model training
- 3Used an image generator to create test images with hidden Easter eggs in various scenes
Details
The author wanted to build a fun app that could detect the presence of Easter eggs in images, which maps to the fundamental computer vision problem of locating small, specific objects in complex scenes. They explored using Amazon Rekognition's standard object detection API as well as training a custom model. The app allows users to upload an image and see if any Easter eggs are detected, with bounding boxes and confidence scores. The author generated 37 test images, some with hidden Easter eggs and some without, to evaluate the capabilities of the standard and custom models. The project code is about 200 lines of Python, with 100 lines dedicated to the user interface.
No comments yet
Be the first to comment