https://github.com/who-else-but-arjun/course_project_da241m
This repository contains the source code for the face recognition and enhancement project which was made as a part of the DA241M course at IIT Guwahati
https://github.com/who-else-but-arjun/course_project_da241m
cnn debluring deep-learning image-processing keras python super-resolution tensorflow vgg16
Last synced: about 2 months ago
JSON representation
This repository contains the source code for the face recognition and enhancement project which was made as a part of the DA241M course at IIT Guwahati
- Host: GitHub
- URL: https://github.com/who-else-but-arjun/course_project_da241m
- Owner: who-else-but-arjun
- Created: 2024-10-21T22:16:54.000Z (7 months ago)
- Default Branch: main
- Last Pushed: 2024-11-05T14:53:21.000Z (7 months ago)
- Last Synced: 2025-02-14T14:57:40.915Z (3 months ago)
- Topics: cnn, debluring, deep-learning, image-processing, keras, python, super-resolution, tensorflow, vgg16
- Language: Jupyter Notebook
- Homepage:
- Size: 1.43 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Face Recognition Attendance System
This project implements a face recognition-based attendance system using several machine learning models. The entire pipeline leverages **SRCNN**, **DeblurGANv2**, and **LIME** models for image processing, along with a custom-trained **VGGFace model** for facial recognition. Below is a breakdown of the files, usage, and instructions for setting up the project.
## Project Structure
### 1. **SRCNN.py**
- **Purpose:** Defines the SRCNN model used for image super-resolution.
- **Details:** Loads the pre-trained weights to enhance image quality.
### 2. **DeblurGANv2.py**
- **Purpose:** Implements the DeblurGANv2 model to remove blur from images.
- **Details:** Loads the necessary pre-trained weights for deblurring.
### 3. **layer_utils.py**
- **Purpose:** Contains essential utility functions and building blocks.
- **Details:** Includes implementations for instance normalization and custom layer definitions required for the models.
### 4. **LIME.py**
- **Purpose:** Implements the LIME model for light enhancement to improve lighting conditions in images.
### 5. **pipeline.py**
- **Purpose:** Cascades the **SRCNN**, **DeblurGANv2**, and **LIME** models to form the complete image processing pipeline.
### 6. **crop.py**
- **Purpose:** Uses OpenCV’s Haar Cascade classifier to crop faces from images.
- **Usage:** Processes raw images from the **Dataset** folder, extracts faces, and saves them to the **Headsets** folder.
### 7. **test.py**
- **Purpose:** Script to test the functionality of the trained VGGFace model.
### 8. **VGGface_VGG16.ipynb**
- **Purpose:** Jupyter notebook used for training the **VGGFace** model and creating the dataset.
- **Dataset Structure:**
- Place images in the format:
```
Dataset/{student_name}/images
```
- The `crop.py` extracts faces from these images and stores them in:
```
Headsets/{student_name}/images
```
### 9. **Attendance.py**
- **Purpose:** Main script for recognizing faces using the trained **VGGFace model** and logging attendance.
- **Details:** Uses the webcam to capture real-time images and recognizes students based on the trained VGGFace model.
### 10. **Class.json**
- **Purpose:** Holds the details of each student in the class as a JSON object, which is referenced during attendance logging.
---
## Prerequisites
- **Python Version:** 3.11.9
- **TensorFlow Version:** 2.17
- Ensure the following folders exist and contain the appropriate files:
- **Dataset/{student_name}/images** (Raw Images)
- **Headsets/{student_name}/images** (Extracted Faces)
---
## Setup Instructions
1. **Install Dependencies:**
Install the required Python libraries using:
```bash
pip install tensorflow opencv-python-headless numpy matplotlib
```
2. **Download Weights:**
Download the pre-trained weights for all models from the following link:
[Google Drive - Weights](https://drive.google.com/drive/u/1/folders/1fHatTQSRryGusJ4VUg5E3R599cmTEtlQ)
Place the downloaded weights in the same directory as the code files.
3. **Train the VGGFace Model:**
Open `VGGface_VGG16.ipynb` and run the notebook to train the VGGFace model on the dataset.
Ensure the dataset follows the structure described above.
4. **Extract Faces for Training:**
Run the `crop.py` script to extract faces from the dataset:
```bash
python crop.py
```
5. **Run the Attendance System:**
Use `Attendance.py` to launch the attendance logging system:
```bash
python Attendance.py
```
---
## Usage
1. **Testing the VGGFace Model:**
Use `test.py` to verify the VGGFace model's performance:
```bash
python test.py
```
2. **Image Enhancement Pipeline:**
Run the `pipeline.py` to process an image through the SRCNN, DeblurGANv2, and LIME models:
```bash
python pipeline.py
```
---
## Folder Structure
```
/project-root
│
├── Dataset/
│ └── {student_name}/images/ (Raw student images)
├── Headsets/
│ └── {student_name}/images/ (Cropped faces)
├── SRCNN.py
├── DeblurGANv2.py
├── layer_utils.py
├── LIME.py
├── pipeline.py
├── crop.py
├── test.py
├── VGGface_VGG16.ipynb
├── Attendance.py
└── Class.json
```
---
## Notes
- Ensure all the pre-trained weights are in the correct folder to avoid loading errors.
- Verify that your webcam is properly connected and recognized by the system before running `Attendance.py`.
- Adjust the model parameters if needed during training in `VGGface_VGG16.ipynb`.
---