Feed India

We care for "Every bite, Every story, Every nourished life"

Comment

Our Idea

Just having a meal isn't enough, taking the right meal for your body and making good nutritional choices is important. With " Feed India " we aim to improve India's Hunger Index ranking and reduce child wastage and stunting by making people aware of their nutrition intake. Through our app, we can start making informed choices, take nutritionist suggestions, bridge the gap in your diet, and nourish your body with the nutrients it deserves.

Motivation :

In the latest Hunger Index Report India was categorised with severity of hunger as serious. While the rate of undernourishment in India stood at 16.6 percent and under-five mortality at 3.1 percent, the prevalence of anemia in women aged between 15 and 24 years stood at 58.1 percent. Though this report has been rejected by GOI, child undernourishment and ignorance of food nutritional values and proper diet planning of pregnant women is still widely prevailing in our country. Thus we came up with the Feed India project to solve this crisis.

Our Features

Calorie Snap

Capture and analyze your favorite Indian food photos to determine calorie counts and discover similar foods.

capture image

Nutri Tracker

Plan and track your diet, monitor your nutritional intake of proteins, vitamins, and other essentials, and make good food choices.

plan

Food Encyclopedia

Explore a comprehensive database of Indian foods, including their ingredients, their taste, and place of origin. Through this, you get to know how rich Indian cuisines are!

food

Pantry Locator

Find the nearest food pantries offering free food assistance based on your City or State.

pantry

Articles

Access informative articles on nutrition, health, and culinary topics to enhance your food knowledge.

article

Nutritionist

Get expert advice from certified nutritionists to support your dietary goals and make healthier choices

nutrition

How we built it

We pre-trained Google's VIT Image Classifier using a dataset from Kaggle having 4k+ Indian food images, then using the model weights and Fast ApI we made an API which can take any image Calorie Snap and detect it. We are passing the results to another api to obtain the food calorie and other values. We similarly collected various nutritional data from various projects and datasets and turned the resources into better visualization using Daisy UI. Nutri tracker enables one to learn the various food values per gram we are taking. Next up, we built the food encyclopedia from Kaggle data and made a server using Fastapi. Also We scraped Niti Aayog to get a list of various ngos based on various Indian cities to build the Pantry tracker. Then using firebase crud operations we have the Articles and further on using static data of various doctors we have a list of nutrionists to contact or book appointment. By clicking on the appointment button, Nutritionist will receive a mail of the patient requesting for a check up.

Challenges we ran into

Due to scraping, one of our teammate's IP got temporarily blocked and then we had to re-run the script from different other IPs. Majorly we faced issues in finding a good Indian food dataset, and other nutritional value of various Indian foods. Popular Food101 didn't consist much of Indian foods generally eaten by various Indian households, so we had to resort to others. Firstly we planned to use RCNN and find every food in any Indian Platter but later due to difficulty in coding and training we decided to detect a single major food from any Indian food snap using VIT. The model weights including the checkpoints and biases were huge and thus we were unable to deploy it. Later on, adding to our difficulties, we faced axios error and parsing the image from frontend to backend was a major issue but eventually after several attempts and scrolling multiple answers of Stackoverflow, we made it.

Accomplishments that we're proud of

Finally, after a lot of sleepless nights, we have successfully built a working prototype that solves the crisis/problem to a great extent.

What we learned

We have learned more about testing and building a Deep Learning API along with what we learned about many Image classification model pipelines.

What's next for Feed India

We are planning to take reviews from the HTM judges and our peers to make this project better. Also, we will be trying to contact some of the NGOs and Medical Students to contribute a few articles and share this web app within their community so that more medical professionals join us. If all plans go accordingly we will be planning to raise funds and bring this project to the needy.

Built With

Share this project:

Updates