Computer Vision - Edge Detection & Segmentation with Deep Learning

Overview

This project explores various edge detection and image segmentation techniques, comparing manual kernel approaches with modern deep learning methods. It includes implementations of classical edge detection operators, segmentation methods, and advanced deep learning models for face detection.

Project Goals

• Implement and compare manual kernel-based edge detection methods.
• Apply various image segmentation techniques.
• Utilize advanced deep learning models for accurate face detection and segmentation.

Techniques Included

Edge Detection (Manual Kernels)

• Sobel Operator
• Prewitt Operator
• Scharr Operator
• Laplacian Operator
• Canny Edge Detection (using Scharr Kernel)

Image Segmentation

• HSV Color Space with Thresholding
• SVM & HOG (Histogram of Oriented Gradients)
• K-Means Clustering

Deep Learning Methods

• Face Detection with MTCNN
• Face Detection using OpenCV's CVLIB

How to Run

To execute this project locally:

1. Clone the repository:

   git clone https://github.com/azizzoaib786/computer-vision-edge-detection.git

2. Install the required dependencies:

   pip install -r requirements.txt

3. Run the notebook:

   • Open computer-vision-edge-detection.ipynb in Jupyter Notebook or JupyterLab.
   • Replace the placeholder path in the notebook with your folder path:

     negatives_patches = r'<change_with_your_path>'

   • Execute all cells (Run All).

Requirements

• Python (3.7 or higher recommended)
• OpenCV
• numpy
• matplotlib
• scikit-learn
• scikit-image
• MTCNN
• CVLIB
• TensorFlow or PyTorch
• Jupyter Notebook/JupyterLab

Contributions

Contributions are welcome! Feel free to fork, enhance, and submit a pull request.

Contact

• Email: azizzoaib786@gmail.com

License

This project is licensed under the MIT License.