https://github.com/aryehky/visagecraft
🔍 VisageCraft is a hands-on deep learning toolkit for facial recognition and image classification. Built with MTCNN, VGGFace, and pretrained CNNs like VGG16 & ResNet, it teaches core computer vision workflows—from identity detection to transfer learning—via interactive notebooks. 👤🧠📊
https://github.com/aryehky/visagecraft
ai deep-learning jupyter-notebook keras mtcnn resnet vgg16 vggface
Last synced: 4 months ago
JSON representation
🔍 VisageCraft is a hands-on deep learning toolkit for facial recognition and image classification. Built with MTCNN, VGGFace, and pretrained CNNs like VGG16 & ResNet, it teaches core computer vision workflows—from identity detection to transfer learning—via interactive notebooks. 👤🧠📊
- Host: GitHub
- URL: https://github.com/aryehky/visagecraft
- Owner: aryehky
- Created: 2023-02-21T03:30:05.000Z (over 2 years ago)
- Default Branch: master
- Last Pushed: 2025-05-23T20:52:33.000Z (5 months ago)
- Last Synced: 2025-06-04T22:31:38.592Z (5 months ago)
- Topics: ai, deep-learning, jupyter-notebook, keras, mtcnn, resnet, vgg16, vggface
- Language: Jupyter Notebook
- Homepage:
- Size: 1.63 MB
- Stars: 2
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 🧠 VisageCraft: Deep Vision with Faces & Features
**VisageCraft** is a creative fusion of deep learning tools for mastering two key areas of computer vision:
* **Face detection and identity recognition** powered by MTCNN and VGGFace.
* **Transfer learning for image classification** using pretrained CNNs like VGG16 and ResNet.
---
## 📓 Interactive Notebooks
| Notebook File | Focus Area | Description |
| ----------------------- | ---------------------------- | ------------------------------------------------------------------------------------- |
| `identity_finder.ipynb` | 😎 Face Recognition Pipeline | Detects and verifies identities using MTCNN and VGGFace with preprocessing workflows. |
| `feature_learner.ipynb` | 🧠 CNN Transfer Classifier | Leverages pretrained CNNs to classify new image sets with limited data. |
---
## 🔍 Key Features
### 👤 Facial Recognition
* **MTCNN Detection**: Localizes facial features across lighting, angles, and expressions.
* **VGGFace Embeddings**: Compares and verifies identity robustness.
* **Preprocessing Suite**: Alignment, cropping, and normalization included for optimal inputs.
### 🧠 Transfer Learning with CNNs
* Fine-tunes VGG16, ResNet50 on new image categories.
* Demonstrates fast adaptation with minimal training data.
* Shows how to reuse high-performing feature extractors in new contexts.
---
## 🛠 Technology Stack
* **Python**: Development and scripting
* **TensorFlow / Keras**: Model training and deployment
* **MTCNN**: Face detection library
* **VGGFace**: Embedding model for identity verification
* **OpenCV / Matplotlib**: Visual debugging and data inspection
* **Scikit-learn**: Evaluation metrics and validation
---
## 📁 Project Structure
```
VisageCraft/
├── identity_finder.ipynb # Facial detection and recognition notebook
├── feature_learner.ipynb # CNN transfer learning for image classification
└── README.md # This file
```
---
## 🎯 Why This Project?
**VisageCraft** offers an approachable way to:
* Understand real-world facial recognition pipelines.
* Learn transfer learning best practices with CNNs.
* Explore deep learning concepts through direct, visual experimentation.
Whether you're curious about AI-powered security, personalization apps, or classification tools, this project gives you the groundwork to build your own solutions.