Description

Abstract:

            The face recognition plays a vital role in identifying or verifying the faces of the person in real-time. In most of the developing countries use face recognition for identifying the criminals, track the people, etc. Like that in this project, face recognition is done with OpenCV and deep learning. First collect the data set of the people by taking an image, computing the image and extracting the face embedding model, then training the face embedded model with deep learning algorithm (SVM) to get the label and recognizer model and finally based on the trained model the faces are recognized.

---

Introduction:

          In face recognition, the OpenCV is used. The OpenCV (Open Computer Vision) which helps the computer to understand the vision of things. First, the basic colours are red, blue, green. In that, we can develop a new colour based on the ratio of the basic colour. In OpenCV, the image is read like an array from that image can be processed by computer.

          In deep learning Support Vector Machine, the algorithm is used to train the extracted face embedding, give the output models as label and recognizer in pickle format.

          Finally, the output of the trained model is like a reference for face recognition.

---

Existing model:

          The face detection is most commonly done in Haar cascading frontal face algorithm but we are classifying the faces with a deep learning algorithm.

---

 Proposed System:

          Since the system uses the OpenCV and deep learning it can run in a laptop with a webcam. The captured image is run through extract embedding that computes 128-d face embedding to quantify a face. From that, the output of the extracted face is run in training to obtain the trained model. After that, face recognition is done.

---

Block Diagram:

---

 Block Diagram Description:

          In this block diagram explains how the face recognition work. Frist the input image is given then detect the face is or not after a crop the face from the image. Send the face to deep learning algorithm and extract the face embedding by 128-d facial embedding. At last, it gives the output model.

---

Project Description: 

            In this project, the dataset is collected among the people by creating a dataset program. Take a greater number of images for more acquires. In extract embedding, it collects all images in dataset and runs 128-d facial embedding to give the embedding as output model. Next in training the deep learning algorithm Sample Vector Machine (SVM) use the embedding model output to give label and recognize of a dataset as output model. Then in recognizer, the output model is taken as input to recognize the face in an image or in the video.

---

Library Used:

• Numpy
• Keras
• Open CV
• Scikit learn
• Imutils

---

Conclusion:

          This face recognition with OpenCV and deep learning project give a good acquires on face recognition. This system can be used in offices, schools, colleges, etc.