Scribd
Upload
10 views

Masked Face Detection Algo

Uploaded by

jksdbnjkdvnj234
Copyright
© © All Rights Reserved
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
10 views

Masked Face Detection Algo

Uploaded by

jksdbnjkdvnj234
Copyright
© © All Rights Reserved
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 3

Steps involved in masked face recognition :

1) Face Detection :

**YOLO (You Only Look Once):**

- **Description:** YOLO is an object detection system that divides an image into a


grid and predicts bounding boxes and class probabilities for each grid cell. YOLO
is known for its speed and efficiency and has several versions, with YOLOv3 being
commonly used for face detection.

**Single Shot MultiBox Detector (SSD):**

- **Description:** Initially developed for object detection, SSD is a real-time


face detection model that uses a single deep neural network to predict bounding
boxes and class scores for multiple objects, including faces, at different scales.

**BlazeFace:**

- **Description:** BlazeFace is a lightweight and efficient face detection model


developed by Google. It is designed for real-time face detection on mobile devices
and is known for its speed and accuracy.

**FaceBox

**Description** :

"FaceBoxes" is a face detection model designed for real-time performance on


Central Processing Units (CPUs)

2) Face Feature Extraction models :

**FaceNet:

**Description:** FaceNet is a face recognition system developed by Google


researchers. Introduced in a paper titled "FaceNet: A Unified Embedding for Face
Recognition and Clustering," FaceNet is known for its use of triplet loss during
training, which encourages the network to minimize the distance between similar
faces and maximize the distance between dissimilar ones in the learned feature
space. FaceNet has been influential in popularizing the use of deep metric learning
for face recognition.

**DeepFace:

**Description:** DeepFace is a face recognition system developed by


Facebook's AI Research (FAIR) lab. Introduced in a research paper titled "DeepFace:
Closing the Gap to Human-Level Performance in Face Verification," DeepFace employs
a deep neural network architecture to perform facial recognition tasks. It utilizes
a deep convolutional neural network (CNN) to map facial features into a high-
dimensional space, where similarities between faces can be measured. DeepFace has
achieved notable accuracy and demonstrated competitive performance on benchmark
face recognition datasets.

**OpenFace:**

**Description:** OpenFace is an open-source face recognition toolkit


developed by researchers at Carnegie Mellon University. It provides a set of tools
and models for face recognition tasks, including face detection, facial landmark
localization, and feature extraction. OpenFace employs a deep neural network
architecture and uses pre-trained models to extract facial features. It is designed
to be versatile and can be used for various face-related tasks.

3) Face Recognition model :

**ArcFace:**

**Description:** ArcFace is a face recognition model that focuses on


enhancing the discriminative power of face recognition systems. Introduced in the
paper "ArcFace: Additive Angular Margin Loss for Deep Face Recognition," ArcFace
employs a margin-based softmax loss during training. The margin is introduced to
increase the angular difference between features corresponding to different
identities, thereby improving the model's ability to distinguish between faces.
ArcFace has been successful in achieving state-of-the-art performance in face
recognition tasks.

**CosFace:**

**Description:** CosFace, also known as Large Margin Cosine Loss (LMCL), is


another face recognition model designed to enhance the discriminative power of the
learned features. Introduced in the paper "CosFace: Large Margin Cosine Loss for
Deep Face Recognition," CosFace introduces a cosine margin to the softmax loss
function. This margin encourages an angular separation between features
corresponding to different classes, improving the model's ability to discriminate
between identities.

**DeepFace:**

**Description:** DeepFace is a face recognition system developed by


Facebook's AI Research (FAIR) lab. Introduced in the paper "DeepFace: Closing the
Gap to Human-Level Performance in Face Verification," DeepFace utilizes a deep
neural network architecture for face recognition tasks. It involves a deep
convolutional neural network (CNN) trained on a large dataset of labeled face
images. DeepFace focuses on mapping facial features into a high-dimensional space,
where the similarity between faces can be measured.

4) Mask Detection :

**YOLO (You Only Look Once):**

**Approach:**
- Train a YOLO model using a dataset that includes annotated images of people
with and without masks.
- Adapt YOLO to detect two classes: one for faces and one for masks.
- Annotate the dataset to include bounding boxes for both faces and masks.

**ResNet:**

**Approach:**
- Use a pre-trained ResNet model as a backbone for a mask detection model.
- Fine-tune the ResNet model on a dataset with annotated images of faces with
and without masks.
- Add additional layers for object detection or segmentation, depending on the
desired output.

**EfficientNet:**

**Approach:**
- Similar to ResNet, use a pre-trained EfficientNet model as a feature
extractor.
- Fine-tune the EfficientNet on a mask detection dataset.
- EfficientNet is known for balancing accuracy and efficiency, making it
suitable for mobile and edge devices.

**MobileNet:**

**Approach:**
- Utilize a pre-trained MobileNet model as a lightweight feature extractor.
- Fine-tune the MobileNet model on a dataset for mask detection.
- MobileNet is designed for efficient inference on mobile and edge devices,
making it suitable for real-time applications.

You might also like