BagAlert

🏆 Hackathon Winning Project - Sony x Raspberry Pi Challenge @Manchester HackABot 2025 🏆

1. Inspiration

We have experienced the anxiety of leaving our belongings unattended in public spaces, especially on campus - this is precisely what BagAlert aims to eradicate.

Whether it's a quick dash to a café, running off to the toilet, or simply needing to step away from your table, there's always that lingering worry about the safety of your possessions on campus.

Traditional solutions like carrying everything with you or asking friends (if we speak we're in big trouble 🤫) or strangers to watch your things are inconvenient or unreliable.

BagAlert aims to provide peace of mind and has the potential to provide real security when you need to briefly leave your belongings behind.

What it does

BagAlert is an intelligent monitoring system that acts as your personal belongings guardian. The system uses computer vision and RFID technology to create a seamless security experience:

• Easy Activation: Simply tap your RFID tag or card to activate protection before stepping away from your belongings.
• Smart Object Detection: The system immediately identifies and catalogs all objects in view (laptops, bags, phones, etc.) using computer vision.
• Continuous Monitoring: While you're away, BagAlert constantly analyses the scene, comparing the current position of your belongings with their initial placement.
• Real-time Alerts: If any item is moved or removed, the system immediately detects the disturbance, highlighting the affected object and recording evidence.
• Deactivation on Return: When you return, simply tap your RFID tag again to deactivate the protection.

The system runs on affordable hardware (Raspberry Pi 5, Sony's IMX500 camera module, an RFID reader and an ESP32) and features an intuitive mobile-friendly web interface that displays a real-time camera feed and your item protection status.

How we built it

BagAlert combines hardware and software components to create a full-stack solution:

1. Hardware Layer:

• Raspberry Pi 5 with Sony's IMX500 AI-powered camera module
• ESP32 microcontroller with RFID reader
• RFID tags/student cards for user identification
• Remote broker for MQTT communication between laptop and Pi 5

2. Software Stack:

• Backend: Python (FastAPI) for camera processing and object detection
• Frontend: React with Material UI for the user web interface

Challenges we ran into:

• Configuring the Raspberry Pi x Sony's AI Camera with our RasPi SSH connection to Pi was unstable & hindered our ability to code remotely, since other code editors were weirdly not working on our Pi).
• Communication between the RasPi & RFID was challenging due to the different architectures.
• ESP32 could not connect to the internet, unlike the rest of our hardware. Made cross-communication between our "MQTT subscriber" laptop and the RasPi difficult.

Accomplishments that we are proud of

• Winning the challenge with limited actual working time: Dylan and I (committee members & hackathon supervision volunteers) were put on a 6h inventory duty from 10:30pm-4am to assist other hackers with their hardware challenges.
• Successfully pivoting from a "failed idea" to BagAlert 8 hours in, aka we made a lot ground in little to no time

What we learned:

Interesting parallels between the camera's processing and neuroscience: the human eye pre-processes a lot of information "at source" before relaying it to the brain, similar to the IMX500 camera's in-sensor AI pre-processing before sending information back to the RasPi. This enables faster response times, while increasing data privacy and reducing energy consumption.