Face Recognition - Realtime Recognition

Table of Contents

1. Introduction

2. Requirement

3. Usage

4. Reference

1. Introduction

• Face recognition is a technology capable of matching a human face from a digital image or a video frame against a database of faces, typically employed to authenticate users through ID verification services, works by pinpointing and measuring facial features from a given image.
• Recognize and manipulate faces with Python and its support libraries.
  The project uses MTCNN for detecting faces, then applies a simple alignment for each detected face and feeds those aligned faces into embeddings model (Facenet).
  Finally, a softmax classifier and cosine similarity method was put on top of embedded vectors for classification task.

2. Requirement

• Tensorflow-gpu (2.1.0)
• Python 3.7+
• Keras 2.2.4
• OpenCV-python (4.4.0)
• Pip or anaconda

  Install MTCNN:

  pip install mtcnn (for pip)
  conda install -c conda-forge mtcnn (for conda)

3. Usage

Crawl data:

python src/save_data.py --name (name of save_dir)

Make sure your computer has a webcam

Preprocessing:

• Train embedding model with your dataset, you need to organize dataset as follows:
  Each person needs 4-5 raw photos with many different angle shooting: front, left, right, …

  Face-Recognition
  └───data/
      └───person1/
      |      └───person1_1.jpg
      |          person1_2.jpg
      |          .....
      └───person2/
      |      └───person2_1.jpg
      |          person2_2.jpg
      |          .....
      └───personN/
             └───personN_1.jpg
                 personN_2.jpg
                 .....

  Augment data:

• After having the raw image files, run augment_data.py to augment more images, each person after running will have 100 different images. Augments: rotation_range = 15, brightness_range=[0.4,1.5], horizontal_flip

  python augment_data.py

  Train embedding model:

  python train_embs.py

  After embedding, embedded file will be saved to output / train_embs.pickle

  Train softmax model:

  python train_classify.py

  The number of classes i’m setting here is 37 classes, so change the number of classes that match your dataset

  Enjoy result:

  <img alt="Qries" src="https://github.com/manhminno/Face-Recognition/blob/master/output/1.jpg">

Label is name of saved-dir - box is green, unknown will don’t have label - box is red. Here label is id of person.

• For image recognition:

  python image_recognition.py --path (path to image) --facedetect (use MTCNN to detect face before recognition - yes/no)

  Output will be saved to output/(name_of_img.jpg)

• For video recognition:

  python video_recognition.py --path (path to video) --facemodel (path of facenet weights) --classifymodel (path of classify model weight) --embspath (path of embed dir)

• For stream recognition:

  python stream_recognition.py

  Make sure your computer has a webcam, here i’m setting webcam 0

4. Reference

• Insightface
• Facenet
• MTCNN
• Facenet-keras model (trained by MS-Celeb-1M dataset): Download