IntruderAlert

Inspiration

Car thefts have been at a all time high in Canada and the US. With over 50% more carjacking in Ontario alone since 2023, that is almost 10,000 cars stolen in the year. This problem also hit close to home, as two of our members dealt with car jackings. We understand firsthand the distress and inconvenienced caused by such incidents propelled us to take action. We envisioned a cutting-edge security system that could provide a formidable defense against car thefts.

What it does

Intruder Alert is a 2 factor authentication mobile application that uses artificial intelligence in facial recognition to verify the owner or the driver of the vehicle. This will allow the system to give power from the battery to start up the car.

How we built it

Intruder Alert uses react-native in the front end to interact with a Node.JS backend API server to make update/read/write/delete calls. The calls interact with our AWS cloud database, where we store raw data inside of S3 and use dynamoDB as a relational-database. The data is then used in our facial recognition ML model called Amazon Rekognition. That will send a binary signal to our arduino board (proof of concept) to either allow power from the battery to the light, or to shut off the circuit.

Challenges we ran into

Our team ran into many problems in our journey of creating this application. We first ran into problems when connecting the front-end react native to the back-end Node.Js API servers. The problem came from that the server was hosted in a localhost rather than a local network, which meant that no outside devices other than the one hosting the server could access it. Our second problem came from the Arduino board, where we did not have a Wi-Fi receiver to allow the on/off of the light. Our team also ran into problems with loading data into AWS, because the file that we wanted to upload initially could not be put into AWS. To solve the problem we had to pivot to learning more about data storage and data warehousing, and find better tools for the jobs we wanted to do.

Accomplishments that we're proud of

For many of us this was the first time that we tackled different technologies that were needed for this project, and we were all able to have it connected and working together in a working application. We are extremely proud of each of our members for taking on the challenge of learning new frameworks, languages, programming concepts to complete this important goal.

What we learned

The team learned a lot about cloud computing and it's different uses such as data storage, data warehousing, data pipelining and data processing for working with our machine learning model. The team learned to use Node.Js to setup API servers so that the front-end and back-end could communicate effectively with each other. We also learned the ins-and-outs of embedded programming for the Arduino board.

What's next for IntruderAlert

Even though we only had only had 36 hours to complete a project, we were able to complete all features we wanted for the initial MVP. It is able to recognize a user's face and store it for 2 factor authentication, the user is able to select different vehicles they own, etc. As for the future of Intruder Alert, the immediate next step is to connect our application to an actual car rather than a proof of concept Arduino board. Furthermore, we wanted to add more than one user to a car for 2 factor authentication, and tracking the car with our hardware (GPS).