I-Sense

I-Sense with my little lens your precious assets in a heart beat. I find their cost to insure them if they're lost.

Comment

Inspiration

We were inspired by the recent uproar of air-bnb theft and property destruction, coupled with the increase in natural disasters (California fires, mudslides). We decided to focus on our customer's need for a simple yet powerful approach to label and document their property.

What it does

A simple futuristic mobile application that capitalizes on a full-camera screen to scan our customers' valuable assets. It serves to scan various objects, label them (their make, model, year) and estimate their current market value.

By using I-Sense, our user spends less than 2 seconds on each item to document their assets and make sure they will never be lost again. We aim to anticipate the best claim for our clients in case of natural disasters, theft or even fire by documenting and backing up all their valuable assets.

How we built it

I-Sense is built around TensorFlow for object detection and labeling. It utilizes their powerful offline model to save performance and battery. Our core focus is to serve our customer in a seamless experience. By implementing a 2-seconds timer, all items will be scanned, evaluated and documented in no-time. The backup infrastructure is a dual secure layer, that saves our customers data on Amazon's online servers (S3). This also serves as a portal for insurance companies to interact with their customers, total assets estimation and potentially alleviate future claims.

Challenges we ran into

Starting with implementing an off-the-shelf TensorFlow model knowing that it was initially conceived for a different application. Having a cohesive application, with a simplistic design was also a considerable challenge to minimize any design complexity. We also aimed for accurate labels for a precise classification (with offline capacity)

Accomplishments that we're proud of

Implementing our first TensorFlow on android devices still remains a great achievement, knowing that this is state-of-the-art technology. Our tight bond as a team enabled us to devise a simple application, with a powerful purpose. We are proud to disseminate our vision that might impact other people's lives.

What we learned

We started almost from a simple level of Android and we managed to grow our level together through extensive debugging and online-research (aka googling). Learning how to handle a TensorFlow Model as much as direct communication is the most valuable output of this experience.

What's next for I-Sense

Ideally, we would train our own TensorFlow model for a more accurate classification of our data. This also emphasizes the need for such a wide dataset of potential items for our clients. We are also seeking to invest in a thorough back-end dashboard for insurance companies to be able to handle our customers' data easily.

Built With

Share this project:

Updates