Analysis and comparison of face detection APIs

DOI: 10.31673/2412-4338.2019.043945

Abstract

Today, more and more information is being accumulated in digital form, including media content, which is a growing segment of the Internet, searching for such content is an important task, but it is significantly different from textual information search and involves image recognition. Images recognition is an automatic comparison of images to objects in a class. There are two major issues in this area: classification and identification. The first helps the search engine to understand what type of object is in the media resource. Only correctly having solved this fundamental problem, the computer will be able to distinguish, for example, a dog from a cat. The second allows not only to find the object category, but also to identify it. The article defines the mathematical formulation of the problem, which is important in order to formalize the accuracy of recognition and to determine the procedure for comparing the capabilities of existing algorithms. The results show that existing APIs allow you to largely solve the problem of face recognition in images, and the next important step is to recognize the face of a person on the move, that is, on video content.

Keywords: neural network, convolutional neural network, tensorflow, computer vision, deep learning, convolutional kernel, weight of the neuron, activation function, feature map, pattern recognition.

References
1. Antonio Gulli, Sujit Pal “Deep Learning with Keras” (2017): 25-36. Print
2. Richard Burton, Iffat Zafar, Giounona Tzanidou “Hands‑On Convolutional Neural Networks with TensorFlow” (2018): 35-85. Print 3. Santanu Pattanayak “Pro Deep Learning with TensorFlow: A Mathematical Approach to Advanced Artificial Intelligence” (2017):10-65. Print 4. Tariq Rashid “Make Your Own Neural Network” (2016):14-37. Print
5. CNN in TensorFlow couse – official site. [Electronic resource]. – Access mode: https://www.coursera.org/learn/convolutional-neural-networks-tensorflow/
6. Face Recognitionarticle.– [Electronic resource]. – Access mode: https://towardsdatascience.com/cutting-edge-face-recognition-is-complicated-these-spreadsheets-make-it-easier-e7864dbf0e1a/
7. Image augmentation tutorial – [Electronic resource] – Access mode: https://machinelearningmastery.com/image augmentation-deep-learning-keras/
8. Math method recognition patterns – [Electronic resource] – Access mode: http://window.edu.ru/resource/800/73800/files/lect_Lepskiy_Bronevich_pass.pdf
9. Keras documentation – official site. [Electronic resource]. – Access mode: https://keras.io/documentation/
10. Python Manual – official site. [Electronic resource]. – Access mode: https://www.python.org/
11. TensorFlow documentation – official site. [Electronic resource]. – Access mode: https://www.tensorflow.org/
12. Transfer learning documentation – [Electronic resource] – Access mode: https://towardsdatascience.com/a-comprehensive-hands-on-guide-to-transfer-learning-with-real-world-applications-in-deep-learning-212bf3b2f27a/
13. VGG16 documentation – [Electronic resource] – Access mode: https://neurohive.io/ru/vidy-nejrosetej/vgg16-model/