After being implemented in image and video processing, the deep learning gave way to a whole practical deployment aimed at object detection and recognition, giving acceptable results in terms of accuracy, allowing the development of tools aimed at calculating the number of people in a crowd.
In this sense, a team of scientists belonging to the Japan Advanced Institute of Science and Technology (JAIST) carried out the development of a new technology whose effectiveness made it possible to obtain a more precise estimate of the density of objects.
The team that developed this technology points out that it can be applied to calculate human density in a public place or vehicle density on a road in order to establish strategies that promote improved public safety and traffic efficiency.
When it comes to obtaining information in public spaces or vehicular traffic lanes, video surveillance is one of the most effective measures.
Through video surveillance, the personnel in charge can obtain data about the number of people or vehicles that transit a given area, as well as associated behaviours and events that can help make improvements in aspects such as safety, security and efficiency.
This process is also known as "crowd counting" where the JAIST research group led by Dr. Sooksatra and Professor Atsuo Yoshikata in conjunction with the SIIT research group in Thailand dedicated efforts to improve their effectiveness through a network that achieved greater performance in the technique used.