Inspiration
To build a lightweight, privacy-focused computer vision solution that converts raw video feeds into actionable retail dwell-time analytics.
What it does
Tracks customer dwell time and spatial coordinates using YOLOv8 and ByteTrack, syncing data in real-time to AWS DynamoDB (eu-north-1).
How we built it
Developed with Python using OpenCV for frame processing, YOLOv8/ByteTrack for object tracking, and boto3 for cloud integration.
Challenges we ran into
Maintaining smooth spatial tracking across video frames and resolving AWS region cloud synchronization hiccups.
Accomplishments that we're proud of
Successfully bridging a computer vision tracking pipeline with a real-time cloud database.
What we learned
How tricky bounding box coordinates can be and how to effectively connect AI models to cloud storage.
What's next for Worasy
Expanding multi-zone heatmaps and adding advanced analytics dashboards for retail stores.
Built With
- ai
- amazon-web-services
- boto3
- bytetrack
- cloud-analytics
- computer-vision
- dynamodb
- machine-learning
- object-tracking
- opencv
- python
- retail-analytics
- yolov8
