Ahaar
Know Your Plate, Shape Your Health🍉
The problem Ahaar solves
Why Ahaar Matters, Especially in India
In India, lifestyle diseases like diabetes, obesity, and heart disease are becoming increasingly common. Around 8-10% of Indians suffer from diabetes, and 40% are overweight. These numbers aren’t just statistics—they represent people in our communities. Ahaar is here to help individuals take control of their health by understanding the nutritional content of their meals.
Ahaar is an easy-to-use app that allows users to take a photo of their meal and instantly get detailed information about its nutritional value—proteins, carbs, fats, vitamins, and more. It’s not just about counting calories; it’s about empowering users to make healthier food choices every day.
Why Ahaar is Crucial for India
Rising Health Issues
With increasing cases of diabetes, heart disease, and obesity, tracking nutrition is more important than ever. Ahaar helps users monitor their food intake and take preventive steps to reduce health risks.
Awareness About Indian Foods
Many Indians eat traditional foods like roti, dal, and samosas without knowing their nutritional value. Ahaar helps users understand the impact of these foods on their health, guiding them to make better choices.
Filling the Knowledge Gap
In India, most people lack detailed knowledge of nutrition. Ahaar bridges that gap by providing instant nutritional data for a wide range of Indian foods, helping users make informed dietary decisions.
Saves Time
Ahaar simplifies the process of meal tracking, providing immediate nutritional breakdowns, saving users time and effort in managing their diet.
Ahaar’s Impact
Ahaar helps prevent health issues by offering easy access to personalized nutritional insights. Whether you're in a bustling city or a rural area, Ahaar empowers you to make better food choices and build healthier habits for the future.
Challenges we ran into
One of the main challenges we encountered while building the Ahaar model was achieving a high enough accuracy for recognizing Indian foods. Many traditional Indian dishes look quite similar, making it difficult for the model to distinguish between them accurately. For example, dishes like dal and chana, or sambar and rasam, have similar appearances but very different nutritional profiles. This similarity led to frequent misclassifications and low model accuracy.
Solution: Leveraging the Gemini 1.5 Flash Model
After multiple attempts at training and fine-tuning with our custom dataset didn’t yield the desired accuracy, we decided to try a more advanced approach. We incorporated the Gemini 1.5 Flash version, which provided enhanced capabilities for recognizing intricate details in images. This model's advanced architecture allowed us to handle the nuanced differences in Indian dishes more effectively, resulting in a significant boost in recognition accuracy.
With Gemini 1.5 Flash, we finally reached an accuracy level that met our standards, enabling Ahaar to reliably distinguish between similar-looking foods and provide accurate nutritional information to users.
Tracks Applied (3)
Best Use of MongoDB Atlas
Major League Hacking
Best Use of Auth0
Major League Hacking
Best Use of Streamlit
Major League Hacking