Gaize | Devpost

Inspiration

One of our teammate's friend was involved in a minor accident as he was distracted by his phone while driving. Our team, saddened, was determined to create something that could tackle this head on. With the transition to self-driving cars, affordable technology that improves driving habits is crucial. Our software is aimed to do just that. It pings the driver he/she is looking outside a certain field of view for a certain amount of time.

What it does

Currently, our website showcases the use of our software by giving the users points based on the efficiency of their driving. Firstly, the user needs to train the software by visiting the Train tab and following the instructions there. They can then check their driving efficiency in the Implementation tab.

How we built it

We built Gaize by making an eye tracking algorithm using javascript. We combined various machine learning algorithms to track eye movements using Tensorflow.js. Our algorithm is capable of adapting to different users as it is calibrated before using. We created a web interface, to showcase a proof of concept, using HTML and javascript.

Challenges we ran into

We had a hard time collaborating with our teammates as we were in different time zones. Gaze detection, as a technology, has huge potential - in our opinion. We started off thinking about how this could be used to find out the bias in a person but later switched to the driving problem because of time constraints. We thought this was a major setback for our team during the hackathon.

Learnings that we're proud of

We are proud of teaming up and creating a working project within a span of two days. It was the first time using tensorflow.js for us and getting it to work was a big deal for us. Sharing of ideas, and collaborating with thought and code were huge challenges for us, and we thought we learnt a lot on the soft skills side during HackDavis 2021.

What's next for Gaize

Our website showcases one of many projects that Gaize can provide. On a broader scale, Gaize can be implemented in many fields. One of the most impactful one being "bias detection". Similar to our vehicle video, a video can be played and the "attention" of a user on the video subject could speak a lot about their biases. Other fields that we look forward to work is surveys. We can create better survey experiences with better results using the data of how much a user is paying attention while taking a survey. Apart from that, we think that customers are a huge part of what a company is and knowing what the customers like would benifit the company. Thus we propose to use this in webpages in order to summarize what the users are more likely to pay attention to. What products attract the customers. Next we think that due to the current pandemic, every student attends classes online and due to the lack of visual feedback for the instructor, Gaize would help the instructors know what grabs the attention of students the most. We think these are great potential projects on gaze detection technology - and our project suffices as a proof of concept for the same.