Inspiration
We are inspired by the GDSC's challenge 'Time-Travel for Good'. We aim to help be the eyes for the visually impaired with the use of Intel Geti, TensorFlow and Opencv.
What it does
The app allows users to capture an image of their surroundings and be informed about both obstacles, such as stairs, and obstructions, such as doors or walls, in their path. This helps them navigate safely, regardless of any impairments they may have.
How we built it
For the front end, we utilised ReactJs and Tailwind CSS, and for the back end, we used Django. To detect the object and classify it as either an obstruction or obstacle, we used Intel Geti.
Challenges we ran into
One of the key challenges we faced was the long training time required to process our dataset. To overcome this, we had to make the decision to reduce the size of the dataset, which involved selecting a smaller subset of data that still maintained enough diversity for effective model training.
Accomplishments that we're proud of
We are particularly proud of the fact that our model achieved a solid accuracy rate of 75%. This accomplishment is rewarding considering the significant amount of time and effort we dedicated to annotating and training the dataset.
What we learned
As a team with limited experience using AI software, we faced a steep learning curve. However, this experience provided us with invaluable insights into working with datasets and the complexities of model training.
What's next for Helping People Travel Through Time
We aim to develop an app that enhances performance and user experience. Additionally, we plan to integrate our app with wearable technology, such as smart glasses, to provide users with an even more intuitive way to navigate their surroundings.
Log in or sign up for Devpost to join the conversation.