Inspiration
I heard about an incident where a jewellery store in Uttarakhand got broken into. The security cameras recorded everything, but nobody noticed the door was left slightly open for hours. That tiny difference went unnoticed, leading to the robbery. That got us thinking, what if there was a system that never blinks and instantly spots these visual changes? That’s how BINGO was born: a vision that never blinks!
What it does
BINGO automatically compares time‑series or sequential images, detects differences, highlights them visually, and notifies users in real time. It acts like an “eyes‑open 24/7” inspection assistant, capable of tracking subtle visual shifts that is perfect for real-life scenarios where human inspection can’t keep up.
How we built it
We built BINGO using Python, Flask, and OpenCV as the core of our image processing pipeline. The backend handles image uploads, runs comparisons, and logs results to a SQL database. A lightweight HTML, CSS, and JavaScript frontend displays live results. WebSockets power the real‑time updates. Pandas helps analyze time‑series insights from logs. Versioning and collaboration were managed on GitHub.
Challenges we ran into
- Calibrating image detection to work with different lighting and camera positions.
- Managing image comparison performance for high‑resolution photos.
- Linking asynchronous image uploads and real‑time WebSocket updates smoothly.
- Deploying the whole project on Azure cloud.
Accomplishments that we're proud of
- Built a fully working prototype with automated image comparison and real‑time visualization.
- Achieved precise detection accuracy even on low‑quality or inconsistent images.
- Turned a simple concept into a deployable cloud app in a short project sprint.
What we learned
We learned a lot about computer vision, how to structure asynchronous Flask APIs, how WebSockets really work under load, and the importance of small pre‑processing steps like image alignment. We also picked up better habits for version control, API design, and balancing backend efficiency with frontend feedback loops.
What's next for BINGO
- Add a mobile‑first version for quick on‑site checks through snapshots.
- Integrate AI models to classify the type of detected change (crack, color fade, movement).
- Build out a reporting dashboard with trends and anomaly analytics.
- Introduce role‑based monitoring for teams using it across different labs or projects.
Log in or sign up for Devpost to join the conversation.