Face detection And Recognition with tensor flow and FaceNet

In the New Year’s, face recognition applications have been created on a lot bigger scope. Face detection and recognition have developed and are being utilized at various spots. The demand for face detection and recognition systems is increasing nowadays quickly. It can be your office’s attendance system or a simple face detector with your mobile’s camera, face recognition systems are everywhere.

Demand for face detection and recognition is even more as they serve many purposes like surveillance of a crowd, passengers at an airport, bus stand, and so on.

What is face detection?

Face detection is aAI based system with which you can detect faces. It can also detect the various part of the face which is called as face landmarks. Here Landmarks are the parts of the human face just like Ears, Cheeks, Eyes, Mouth, and Nose, etc. Face detection plays a big role in developing face tracking, face analysis, and facial recognition.

What is Face Recognition?

Although the terms face detection and face recognition are generally used together. But both are two different things as Face recognition is one of the important applications of face detection. Face detection means recognizing the detected face. We can say that Face recognition extends the functionality and of detecting human face and include the feature of determining who it is.

 

Machine Learning Models 

  • FaceNet
  • TensorFlow Lite Model

 

FaceNet :

FaceNetwas published in 2015 via Google researchers Schroff et al.FaceNet is a deep neural community used for extracting features from a picture of a person’s face. FaceNet is a face recognition model that learns the mapping from faces to a location in a multidimensional space where the distance between points directly corresponds to a measure of face similarity.

Face Net System can be used to get high-quality functions or specifications from faces, called embedding, which can be used to train a face identification model.

The mode is a deep convolutional neural network trained via a triplet loss function the improve vectors for the same identity to be more similar, The focus on the training model was important in this work.

 

Tensor Flow Lite Model:

When A machine learning model is implemented and gets trained with TensorFlow, normally you will end up with a model file that needs a sample storage space and also a Graphical interface to run But as the large storage space and GPU are not can be performed on all mobile phones.

“Tensor Flow Lite comes with a solution to enabling Machine Learning Models to run on mobile devices.”

It provides a set of tools that enables on-device machine learning by helping developers run their models on mobile, embedded, and IoT devices.

On Device Inference: this refers to the capability of utilizing the TensorFlow lite model within the mobile device and more importantly, devices with a constraint on storage space.

 

Face Recognition Solution with Three Stages:

  1. Pre-Processing :
  • In this stage, it takes a set of images and converts them all into a uniform format
  1. Embedding :
  • A Process, basic to the way facets work, learns a representation of faces in multidimensional space at where distance corresponds to a measure of face similarity.
  1. Classification: The Last Stage uses information provided by the embedding process to separate distinct faces.

Recent Blogs

Mr. Nisarg Shah
Building an Audio & Video Call App with…
December 11, 2020
What exactly is outsourcing in business, and especially in the tech industry? Modern companies are faced with uncertainty as now, new trends and technologies arise and change fast. To keep up with the trends..
Explore More
Mr. Nisarg Shah
IT Outsourcing Strategies: Types, Pros, and Cons of…
December 11, 2020
What exactly is outsourcing in business, and especially in the tech industry? Modern companies are faced with uncertainty as now, new trends and technologies arise and change fast. To keep up with the trends..
Explore More