Article Preview
Top1. Introduction And Problems
The Corona pandemic suddenly came to the world, while the world was not prepared for it, even with the slightest precautions, which led to many human losses. When we talk about the world, we highlight the Arab world, especially Algeria, where there is a serious lack of artificial intelligence technology and computer vision.
The goal Artificial Intelligence (AI) is to create systems that can function intelligently and independently in order to make the machine to mimic human consciousness. The ideal feature of AI is its ability to make a decision. The techniques from the field of AI, and more specifically Deep Learning (DL) methods, have been the core components of more recent developments in the field of computer vision, where it was exploited to solve the biological problems and diseases such as COVID-19.
Although there is a vaccine for the Coronavirus, it is still spreading, currencies, the impact of COVID-19 is widespread and has broad implications; it can be broken down different social domains such as economy, healthcare and social services.
We came up with the idea that the vaccine cannot eliminate the coronavirus unless it goes in parallel with taking precautions (codified by the World Health Organization (WHO)) for any future pandemics. The idea behind the project is to reduce coronavirus disease by tracking people using surveillance cameras, sensors, and drones to discover the wrongs committed in citizens such as not wearing a mask or not respecting social distancing.
1.1 Aims and Objectives
- •
Develop a powerful system to stop the spread of covid - 19 especially and facilitates the process to eliminate any problem of this pandemic in the future.
- •
The proposition of a new version of MobileNet model and YOLO to detect social distancing violation and mask detection.
- •
Create an IOT environment to help police officers and controllers to track people in abnormal situations of covid-19 through their mobile and smartwatch which will connect with drones, surveillance cameras and sensors. This is why we have chosen mobilenet architecture.
- •
Face mask detection and social distancing using new proposed models and configuration by changing the DarkNet of YOLO and MobileNet architectures.
- •
Analyzing the impact of different techniques of data augmentation in the final results.
- •
The use of transfer learning to accelerate training and achieving more accurate results with optimizing data and time.
2. Review Of Literature
This section provides an overview of CNN models that touches three aspects: classification, object detection and object tracking around two problems: mask detection and social distancing detection.