https://github.com/codewitheshayoutube/imageembedder_using_deeplearning-1
https://github.com/codewitheshayoutube/imageembedder_using_deeplearning-1
Last synced: 2 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/codewitheshayoutube/imageembedder_using_deeplearning-1
- Owner: codewithEshaYoutube
- License: apache-2.0
- Created: 2025-03-12T08:48:20.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-03-12T08:57:26.000Z (over 1 year ago)
- Last Synced: 2025-03-12T09:39:04.631Z (over 1 year ago)
- Language: Python
- Size: 0 Bytes
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
- Codeowners: .github/CODEOWNERS
Awesome Lists containing this project
README
# 📌 ImageEmbedder: AI-Powered Object Embedding
[](https://image-embedder-using-deeplearning.streamlit.app/)
## 🌟 Overview
**ImageEmbedder** is a Streamlit-based web app that allows users to seamlessly embed objects into images using AI. It utilizes **CLIP (ViT-L/14) from Hugging Face** to intelligently determine the **best placement** for objects based on semantic similarity.
## 🛠 Features
✅ Upload a **background image** and an **object image**
✅ Uses **CLIP model** to find the **optimal placement** for the object
✅ Supports **manual adjustments** (X, Y position & scaling)
✅ **Seamless blending** using OpenCV & PIL
✅ Web-based interface using **Streamlit**
---
## 🚀 Technologies Used
- **Python 3.x**
- **Hugging Face Transformers** (CLIP model: `openai/clip-vit-large-patch14`)
- **OpenCV** (for image processing)
- **NumPy** (for numerical operations)
- **PIL (Pillow)** (image manipulation)
- **Streamlit** (for UI)
---
## 📥 Installation
### 1️⃣ Clone the Repository
```sh
git clone https://github.com/yourusername/ImageEmbedder.git
cd ImageEmbedder
```
### 2️⃣ Install Dependencies
```sh
pip install -r requirements.txt
```
---
## 🎯 Usage
### 🔹 Run the Streamlit App
```sh
streamlit run app.py
```
### 🔹 Steps
1. **Upload a background image**
2. **Upload an object image (PNG with transparency recommended)**
3. Click **"Find Best Placement with CLIP"**
4. AI suggests the **best position** based on context
5. Optionally **adjust manually** & save the final image
---
## 🔬 How CLIP is Used
1. **Extracts features** from the background & object images.
2. **Computes similarity scores** across different regions.
3. **Finds the most contextually relevant area** to place the object.
4. **Blends the object** seamlessly into the background.
---
## 📌 Example Output
---
## 📌 Future Enhancements
🚀 Add **text-based placement** (e.g., "Place near the tree")
🚀 Improve **adaptive scaling** for different image sizes
🚀 Support **multiple object embeddings**
---
## 📜 License
This project is **open-source** under the MIT License.
👨💻 Built with ❤️ by **Eesha Tariq** during Research Internship at **University of Sharjah**