🍳 DeepChef

DeepChef is an intelligent recipe recognition web app that identifies dishes from photos and suggests their recipes. Upload an image of a dish, and DeepChef will recognize it using deep learning and return the top 5 most similar recipes!

---

🚀 Features

• Image Upload Interface: Click and upload any food image.

• Dish Recognition: Uses a pre-trained DenseNet (Keras) model to extract image features.

• Recipe Retrieval: Finds top 5 matching dishes using cosine similarity (arccosine distance) between image feature vectors.

• Dataset:

  • Scraped 380+ recipes from food.com.
  • Collected 10 images per dish via Google Images using Selenium.

• Web App: Built using Django, HTML, CSS, and JavaScript.

---

🧠 Tech Stack

• Backend: Django
• Frontend: HTML, CSS, JavaScript
• Deep Learning: Keras with DenseNet
• Web Scraping: Selenium, BeautifulSoup

---

⚙️ How do we use it?

1. User uploads an image of a dish.
2. The image is passed through DenseNet to generate a feature vector.
3. These features are compared (via arccosine similarity) with precomputed features of the dataset.
4. The top 5 most similar dishes are displayed along with their recipes.