Inspiration
As someone who goes to the gym regularly, I always knew that nutrition is 80% of the battle. I set distinct daily nutritional targets—hitting 150g of protein while keeping added sugars under 35g. However, my biggest hurdle was never the workout itself; it was managing my food, kitchen, and tracking habits. Every existing fitness app forced me into the same tedious cycle: manually searching databases, weighing individual ingredients, and typing in item by item. After an exhausting leg day, the last thing I wanted to do was log food for 10 minutes. To save time and money, I tried meal prepping large batches on Sundays. But reality would catch up by midweek. I’d get home exhausted from the gym, forget what was sitting in the back of my fridge, give in to food delivery, and end up throwing away spoiled chicken curry by Friday. I realized two major problems were feeding into each other: diet tracking fatigue and household food waste. I wanted to build an app that solved both at once—connecting post-workout fitness goals, real-time macronutrient balancing, meal prep logging, and smart leftover management into a frictionless daily loop.
What it does
NutriScan AI turns quick meal photos into effortless macro tracking while transforming fridge leftovers into zero-waste, personalized meals. 📸 1. AI Photo Calorie & Macro Estimator Snap & Log: Instead of manually searching databases or typing ingredients, users simply snap a photo of their plate. Multimodal Extraction: Powered by Gemini 3.5 Flash, the app parses the image to estimate portion sizes, total calories, and exact macronutrient breakdowns (Protein, Carbs, Fiber, and Sugar).
📊 2. Dynamic Fuel Gauges & Real-Time Counter-Balances Visual Gauges: The home screen dashboard features dynamic progress bars/rings that update instantly as meals are logged. Threshold Alerts: If a user exceeds a daily target (such as added sugars), the progress gauge turns red to display a warning (e.g., SUGAR WARNING: Daily Limit Exceeded). AI Diet Copilot: Rather than just alerting the user, the app provides immediate, actionable recommendations—like suggesting a high-fiber, low-carb swap for dinner to flatten the glucose response.
🧊 3. Smart "Virtual Fridge" (Meal Prep Tracker) Large Batch Toggle: When logging a meal-prepped dish, a simple toggle logs one active serving for the day and automatically saves the remaining portions to a database with a timestamp. Freshness Management: Keeps track of active leftovers in the kitchen to prevent cooked food from being forgotten and going to waste.
🔁 4. Macro-Aware Leftover Transformer Automated Expiry Alerts: When leftovers hit a 2- to 3-day window, background triggers send in-app or push notifications reminding the user to consume the stored meal. Contextual Recipe Generator: Instead of recommending generic recipes, the app queries Gemini to generate a quick, 10-minute recipe (e.g., turning leftover curry into a high-protein wrap) calculated specifically to match the user's remaining daily macros.
🛒 5. Preference-Matched Smart Grocery List Cuisine Customization: The app adapts to user taste preferences (e.g., Indian, Mexican Fiesta). Auto-Populated Ingredients: When selecting a recipe, it cross-references existing pantry items and automatically adds only missing ingredients to an organized grocery list.
How we built it
1.Multimodal Meal Logging: Instead of typing ingredients, I pass meal photos straight to Gemini Vision. The model detects food types, estimates portion sizes, and outputs structured macronutrients (Protein, Sugar, Carbs, Fiber). 2.Dynamic Fuel Gauges & Real-Time Counter-Balances: The dashboard visualizes intake thresholds. If a sugary snack pushes my intake over the limit, the sugar gauge turns red and triggers an actionable AI Diet Copilot recommendation (e.g., suggesting a high-fiber, low-carb dinner to flatten the glucose response). 3.The "Virtual Fridge" Engine: When logging a meal-prepped dish, a simple "Large Batch" toggle logs one active meal portion while saving the extra portions into a Firestore virtual_fridge collection with an automated timestamp. 4.Macro-Aware Leftover Transformer: When leftovers sit for 2–3 days, background dispatch guards issue notifications. Tapping the alert doesn't just tell me to microwave old food; it queries Gemini to generate a fresh, 10-minute recipe matched to my remaining daily macros.
Challenges we ran into
2D Depth Perception for Portion Sizes: Estimating precise calories from a flat 2D photograph is notoriously tough because images lack volumetric depth. I solved this by engineering system prompts that instruct Gemini Vision to cross-reference food density baselines against standard plate sizes. Context-Aware Leftover Recipes: Generic recipe APIs often suggest meals that require buying a long list of brand-new ingredients, defeating the purpose of clearing out the fridge. I crafted system prompts forcing Gemini to treat virtual_fridge items as primary constraints, allowing only minimal pantry staples and keeping prep time under 10 minutes. State Timestamps & Notification Guard Rails: Managing leftover state across days meant preventing repeat alert spam. I implemented a sentReminders date-keyed tracker state that verifies timestamps so users receive a single dispatch prompt at the 48-hour and 72-hour marks.
Accomplishments that we're proud of
We are immensely proud of building an app that seamlessly bridges post-workout nutrition with zero-waste kitchen management. Beyond the tech stack, the real victory came when I gave it to my gym mates and it helped them a lot healthily and financially—it eliminated their diet-tracking fatigue by replacing manual entry with instant Gemini photo parsing , while keeping them accountable to daily protein and sugar targets. By guiding them to rescue batch-cooked leftovers from the Virtual Fridge instead of ordering $15–$20 fast-food delivery on exhausted evenings , NutriScan AI didn't just optimize their post-gym recovery; it saved them real money every week while stopping household food waste in its tracks.
What we learned
Linking gym nutrition targets to kitchen inventory management solves both fitness goals and food waste at the same time.
What's next for NutriScan AI
We would like to let the user add their health reports. Our app customizes the diet plan according to their medical conditions.
Log in or sign up for Devpost to join the conversation.