Skip to content

Latest commit

 

History

History
64 lines (50 loc) · 2.72 KB

README.md

File metadata and controls

64 lines (50 loc) · 2.72 KB

Object-Detection-with-Haar-Cascade

This project uses OpenCV's Haar Cascade for face detection and performs hyperparameter tuning to improve detection accuracy.

Approach

The face detection model used here is a pre-trained Haar Cascade classifier, which operates by scanning the input image in a sliding window approach.

1. Hyperparameter Tuning

To improve detection accuracy, the project employs hyperparameter tuning on parameters such as scaleFactor, minNeighbors, and minSize.

  • scaleFactor : Image scaling factor.
  • minNeighbors : Minimum number of neighbors for a rectangle to retain it.
  • minSize : Minimum size of the bounding box for detection.

The evaluation of the model's performance is based on:

  • True Positives (TP): Correctly detected faces in images with faces.
  • False Positives (FP): Non-face images incorrectly classified as containing faces.
  • True Negatives (TN): Correctly identified non-face images.
  • False Negatives (FN): Faces missed in images containing faces.

The accuracy is calculated as:

$$ \text{Accuracy} = \frac{TP + TN}{TP + TN + FP + FN} $$

The tuning function iterates through different combinations of parameters to find the best configuration for maximizing accuracy.

2. Dataset

This project uses two datasets:

  • Face Dataset: This dataset consists only of faces of humans.
  • Non-Face Dataset: This dataset consists of images of flowers, airplanes, motorcycles, nature, dogs, and cats, which are not considered faces of humans..

Installation

  1. Clone the repository: 
    git clone https://github.com/sahasri/Object-Detection-with-Haar-Cascade.git
cd Object-Detection-with-Haar-Cascade
  2. Install the required dependencies: 
    pip install -r requirements.txt
  3. Download dataset: The required datasets are included in the repository as zip files.
  • faces-dataset.zip: This zip file contains 200 images of faces of human.
  • non-faces_dataset.zip: This zip file contains 200 images of non-faces, such as flowers, airplanes, motorcycles, nature, dogs, and cats.
  • And also datasets are available in kaggle. 
      kaggle datasets download -d sahasrimanimendra/non-faces-dataset
  kaggle datasets download -d sahasrimanimendra/faces-dataset

File Structure

Object-Detection-with-Haar-Cascade/                     
├── faces_dataset/                          
│   ├── 001.jpg
│   ├── 002.jpg
│   └── ...

├── non_faces_dataset/                      
│   ├── 001.jpg
│   ├── 002.jpg
│   └── ...

├── Object-Detection-with-Haar-Cascade.ipynb  
├── requirements.txt                                                     
└── README.md