kasra1377 / mask-detection

An application for detecting mask on faces

😷Mask Detection Project

---

📄Description

---

• In this project our purpose is building a model that detects mask on face(s).
• The model of this project is a pre-trained model that is fine tuned from mobileNetV2 model.
• If you are interested in further details, you can access them in mask-detection-training.ipynb file in this repository

📐Models

---

mask-detector.model: A pre-trained model that is fine tuned from mobileNetV2 model. This model detects whether the face is with mask or without mask.

deploy.prototxt & res10_300x300_ssd_iter_140000.caffemodel: Two essential CV2 DNN models to detect faces in images. So you have to download them to implement this project.

Note : These models are available in this repository.

📈📉📊Model Performance

---

• This model has reached 99% accuracy in validation set and training set in 20 epochs which is acceptable.
• This model classified all the samples correctly, except ~5 of them.
• You can see the classification report and confusion matrix for further details down below.

🎫 Dataset

---

• You can access the dataset via mask_dataset folder in this repository.
• The folder contains two subfolders, with_mask and without_mask.
• without_mask folder contains cropped faces. These faces do not have mask.
• with_mask folder contains persons that has a mask on their faces.
• For avoiding any kind of biases we did not put the same face in these two subfolders.
• The dataset contains two classes, and each of them has ~1000 samples. So in this project we will not face an imbalanced dataset.

💻🖥Installation

---

🛠Requirements

Module/Framework	Version
tensorflow	2.4.1
sci-kit learn	0.22.2.post1
seaborn	0.11.1
pandas	1.1.5
numpy	1.19.5
cv2	4.1.2
PIL	7.1.2
matplotlib	3.2.2
imutils	0.5.4

⚙Setup

---

Tensorflow:

$ pip install tensorflow==2.4.1

Scikit-learn:

$ pip install scikit-learn==0.22.2.post1

Seaborn:

$ pip install seaborn==0.11.1

Pandas:

$ pip install pandas==1.1.5 

Numpy:

$ pip install numpy==1.19.5

CV2:

$ pip install cv2==4.1.2 

PIL:

$ pip install PIL==7.1.2

Matplotlib:

$ pip install matplotlib==3.2.2 

Imutils:

$ pip install imutils==0.5.4

👥Contributers

---

Kasra1377