Inspiration
Nowadays, unlike in past times, children have lost the touch of Grass. Due to the advance of technology and the fast track of the world, they have been parted from the basic of the necessary information like information on common insects, flowers, trees, plants and much more, which does not seem a lot of issue at the small scale but ignoring that issue might create more difficulty in their development. To overcome that, we developed an interactive application that captures images of various objects, detects them, and provides essential information.
What it does
The application performs various AI and ML operations to detect objects from images and, based on that, provides essential information.
How we built it
We started by using various pre-trained models. Further, we fine-tuned the pre-trained model with the appropriate dataset to get more accurate and niche results. In this process, initially, we used Gemini API to detect objects and give information about the image. Further, we pass that information on to our newly fine-tuned model to summarise it. Finally, our application provides the required details.
Challenges we ran into
We faced a lot of challenges mentioned below but not limited to:
Finding the appropriate model Finding the proper dataset for fine-tuning the model Training the datasets of almost 7.7 million data.
Accomplishments that we're proud of
Even though there were many challenges, we tried our best to accomplish it all and deliver our ideas. We are more than happy to discuss the problem that we are solving, which can make a difference in the lives of children and has the future scope of making more difference in other domains.
What we learned
Working on the problem and the issue we were trying to solve has helped us explore and learn a lot on the technical side, like AI, ML, Intel AI Tools, Gemini API, Frontend, Backend, and much more. Still, it's not only limited to the technical side. Apart from these, some of the essential soft skills like teamwork, communication skills, networking, hard work, delivering work on time, and the scope of always having something new to learn are also essential.
What's next for Touch Grass
Having said the above things, Touch Grass can have far more features, not only for children but also, in the future, for teenagers, middle-aged, and older people. The model trained with a specific fine-tuned dataset can result in a more niche AI model that can be used in various sectors.
Log in or sign up for Devpost to join the conversation.