EL TALAGA
is an AI-powered, personalized meal-planning application built with Flutter. It bridges the gap between food waste reduction and daily meal prep by utilizing cutting-edge Computer Vision and Generative AI to tell users exactly what they can cook based only on the ingredients they currently have in their fridge.
# 🎯 The Core Problem It Solves
Every day, people open their fridges, stare at a random assortment of ingredients, and think: "I have nothing to eat." This leads to expensive food delivery, unhealthy eating habits, and massive amounts of food waste. EL TALAGA eliminates mealtime decision fatigue by acting as a personal, AI-driven chef in your pocket.
# ✨ Key Features & User Workflow
1. Smart User Profiling (Onboarding)
Before cooking, the app learns how the user wants to eat. It captures critical constraints that the AI will strictly follow:
Dietary Restrictions: (e.g., Vegan, Keto, Gluten-Free) Allergies: (e.g., Peanuts, Shellfish, Dairy) Dislikes: (e.g., Cilantro, Mushrooms) Serving Sizes: Number of people the recipe should feed.
2. AI Vision Fridge Scanner
Users take or upload a photo of their fridge shelves. The app uses an on-device YOLOv9t Object Detection Model to instantly scan the image. It processes the pixels and identifies raw ingredients (e.g., Tomatoes, Eggs, Milk, Chicken, Rice) with bounding-box accuracy.
3. Ingredient Verification Interface
Since AI vision isn't always perfect, the app presents a clean, editable list of the detected items. The user can easily delete false positives with a swipe or add missing items using a text-search input, ensuring the recipe generator has 100% accurate data.
4. "Chef AI" Recipe Generation
Once the ingredients are confirmed, the data is passed to a powerful Large Language Model (Gemini / Ollama). The LLM is prompted with a strict system persona to design a delicious, practical recipe that:
Prioritizes the exact ingredients from the user's fridge. Strictly avoids any flagged allergies or dislikes. Adheres to the selected dietary plan.
5. Interactive Cooking Mode
The generated recipe isn't just a block of text. The app parses the LLM's markdown output and formats it into a beautiful, interactive UI with three distinct tabs:
Cookware: What pans and utensils to prepare. Ingredients: Exact measurements needed. Instructions: Step-by-step cooking guide. A sticky "Start Cooking" bottom bar prepares the user to enter a hands-free cooking mode.
Built With
- africa's-talking
- c++
- food
- hungry
- kotlin
- water

Log in or sign up for Devpost to join the conversation.