Face Detection Using Computer Vision
p. 1
Nimai Chand Das Adhikari
• Plays an important role in many applications such as video
surveillance and face image database management
- Each person has a particular face structure
- Facial symmetry
• A survey on:
- Different Facial recognition algorithms as well its implementation
- Tests of different feature extraction and dimensionality reduction
methods
A Brief Note
Steps FollowedSteps Followed
Layers:
1. Convolutional Layers:
To extract different features
of the input image with
RELU activations
2. Pooling Layers:
Max operation on the
obtained feature map vector
3. Fully Connected Layers:
To predict the probabilistic
labels
CNN: Convolutional Neural Network
Architecture Used
+ =
Architecture Description
Face DetectionFace Detection Process
Use of Open CV
Steps:
1. Open CV: Computer Vision
Package
2. Detect the Face and Eyes
Results Achieved
• Dataset: Yale Face
Database
• Size: 6.4MB
• # of Images: 165
• 15 individuals
• 11 images/subject
Network Output Graph
Results Output
Dataset Used
p. 8
THANK YOU!!

Face detection Using Computer Vision

  • 1.
    Face Detection UsingComputer Vision p. 1 Nimai Chand Das Adhikari
  • 2.
    • Plays animportant role in many applications such as video surveillance and face image database management - Each person has a particular face structure - Facial symmetry • A survey on: - Different Facial recognition algorithms as well its implementation - Tests of different feature extraction and dimensionality reduction methods A Brief Note
  • 3.
  • 4.
    Layers: 1. Convolutional Layers: Toextract different features of the input image with RELU activations 2. Pooling Layers: Max operation on the obtained feature map vector 3. Fully Connected Layers: To predict the probabilistic labels CNN: Convolutional Neural Network
  • 5.
  • 6.
    Face DetectionFace DetectionProcess Use of Open CV Steps: 1. Open CV: Computer Vision Package 2. Detect the Face and Eyes
  • 7.
    Results Achieved • Dataset:Yale Face Database • Size: 6.4MB • # of Images: 165 • 15 individuals • 11 images/subject Network Output Graph Results Output Dataset Used
  • 8.