AV1 for Intelligent Video Analytics

Want create site? With Free visual composer you can do it easy.

The Intelligent Video Analytics tool developed within ESI offers more efficient management of video data in both direct operation and search and playback. This solution frees operators from the problems of uncontrolled alarm proliferation through the use of Deep Learning technology.
This technology consists of teaching a specific task to a computer model. The tasks learned can be focused on images, sound or text, in other words, all the data that we find in the field of video surveillance and remote surveillance. In recent years, the Deep Learning performance has reached an exceptional level of precision, which is often superior to human performance. These models are a set of artificial neurons that have the ability to continually improve when they are fed with new learning data.
On the other hand, unlike conventional motion detection approaches, Deep Learning methods are not affected by different view angles found in video surveillance, any type of atmospheric pollution (rain, lighting changes, etc.) or animals, meaning that the number of false alarms is very limited.
AV1 automatically ensures the correct positioning of the cameras, the filtering and elimination of false alarms and the detection of behaviour that is normally dealt with by a human operator. AV1 works as a virtual operator, connecting to sites for an automatic video patrol. With this solution, a single operator can monitor a multitude of different sites in one patrol.
AV1 also has features to detect different categories (Person, Car, Dog, Cat, etc.), to recognise human actions (Walking, Running, Raising arms, Lying down, etc.), to determine human attributes (Age, Body size, Gender, etc.) and to identify faces.
Metadata are associated with the analysed images (attributes, categories detected) and will be able to optimise film search and replay in the future. Camera efficiency is also increased because in the space of half a second, there are 999 chances out of 1000 of detecting a person.
This future is constantly improving and the ESI team in charge of Deep Learning is continually enhancing AV1 to optimise operator productivity and make their work more reliable. In addition, it avoids the use of other means of removing doubt and thus limits the additional costs incurred.

Did you find apk for android? You can find new Free Android Games and apps.
Posted in ESI