X-secure | Devpost

App Design
Network Architecture

Inspiration

A lot of armed robbery incidents and thefts happens in shops all around the world. They can be prevented if necessary actions are taken in time. We thought about a ml/ai based system that can help prevent such incidents and make shops secure. There's a range of WIFI enabled cctv camera's available online. So, we thought of using android based advantages it gives us.

What it does

Our project uses ai/ml for emotion ,gesture, pose estimation and detection, predicting the situations. If it detects something unusual it notifies the shop owner immediately and in case of robbery notifies emergency services as well. It also stores such events separately and highlight suspected events.

How we built it

We used open-cv to build emotion, gesture and pose estimation model. Open pose model was used to bulid a 18 point model trained on COCO dataset for human pose estimation.link
We also designed an app for instant notification, contacting emergency services, manage security cameras and video proofs. link

Challenges we ran into

This is our first hackathon in terms of ml/ai. We had a hard time accomplishing the accuracy required for optimal performance of the model. We had no prior experience in android ui design.

Accomplishments that we're proud of

We are proud that we were able to complete our first ml/ai hackathon in such a short time and deliver with some quality response. We are also proud of the new skills we grasped.

What we learned

We had no previous experiencing in designing and developing Android app UI using Figma or Open Pose Model. So we had a chance to learn new tech like cnn, opencv, ui design and enhance our skill set.