In today’s health-conscious world, understanding what we eat is crucial for maintaining a balanced lifestyle. Imagine an app that can analyze your meals just by looking at a photo — providing detailed calorie counts, nutritional breakdowns, and health assessments instantly. This exactly helps to build by leveraging the power of Google Gemini Pro Vision API.

The Vision: Empowering Personal Nutrition Through AI

Motivated by personal health goals and a passion for balanced nutrition, developed a food calorie and nutrition analysis app from scratch. The app enables users to upload images of their meals and receive comprehensive insights about calorie content, individual food items, and overall meal healthiness.
This Blog showcases how artificial intelligence, particularly computer vision combined with prompt engineering, can transform how we approach nutrition.

Core Features for Building a Smart Food Calorie & Nutrition App

The app offers an intuitive, seamless experience built around these key functions:

1. Image Upload & Display: Users simply upload a photo of their meal (JPG, JPEG or PNG), which is immediately shown on the screen for confirmation.
2. Smart API Integration: By integrating with Google Gemini Pro Vision API, the app converts the uploaded image into the required data format and sends it to the API alongside custom-crafted prompts.
3. Prompt Engineering for Expert Analysis: The API is guided with prompts that instruct it to behave like a nutrition expert. These prompts request identification of food items, calorie calculations per item, total calories, and a health assessment including nutrient ratios such as carbohydrates, fats, fibers, and sugars.
4. Parsing & Displaying Results: The app parses the API’s response to extract meaningful data — itemized calories, nutritional breakdown, and health insights — and presents them in a user-friendly, natural language structured format.
5. Health Recommendations: The app evaluates meal healthiness and suggests improvements to encourage balanced diets rich in proteins, fibers, and essential nutrients.

Behind the Scenes: Technical Walkthrough

Developing this AI-powered nutrition app involves below orchestrated steps:

Sample Code Highlights

Here are some snippets that illustrate key parts of the app:

API Initialization

import streamlit as st
import google.generativeai as gen
import os
from dotenv import load_dotenv

load_dotenv()
gen.configure(api_key=os.getenv("GOOGLE_API_KEY"))

Image Upload & Preparation

uploaded_file = st.file_uploader("Choose an image", type=["jpg", "png"])
if uploaded_file:
    image_bytes = uploaded_file.read()
    image_data = {"mime_type": uploaded_file.type, "data": image_bytes}

Making API Calls

def get_nutrition_response(prompt, image_part):
    model = gen.Model("gini Pro Vision")
    response = model.generate_content(prompt=prompt, image=image_part)
    return response.text

UI Flow

if st.button("Analyze Calories"):
    prompt = "You are an expert in nutrition. Analyze the food items in the image..."
    result = get_nutrition_response(prompt, image_data)
    st.write(result)

Output Screenshots:

By entering Input Prompt and Uploading a meal image (.png, .jpg, .jpeg), will get an estimation of total calories in the meal.

Final Thoughts

As health tech continues to evolve, combining AI-driven image recognition with nutritional expertise promises to revolutionize personal wellness. This journey in building this calories and nutrition app shows how accessible and impactful such technology can be.

This would help to start building your own smart health tools today.

Neha Vittal Annam