Mats Thulin summarized the conditions

[rakhirhif8963]

New technology needs to be tested and adapted before it can be applied to practical applications. “This is where we are today in the field of deep learning analytics for video surveillance, and for this reason we have taken a cautious approach to implementing this technology in our products and applications,” said Mats Thulin, Director of Core Technologies in R&D at Axis Communications.

He noted that in the surveillance field, deep learning is still in the early stages of high expectations. However, there are already areas where it is useful, such as finding objects or events in a huge volume of descriptions.

for applying AI in the observational field: “We can use it croatia whatsapp data the scope of application is limited, when we have large amounts of training data, when there are clear and easily identifiable labels for data annotations, and finally when we know the application scenario and understand the balance between positive and negative false positives.”

In addition, the relevance of the data is important, since the quality of the deep learning algorithm cannot be better than the quality of the original data. The ability of cameras to process images also plays a big role. After all, such solutions must work constantly, cope with complex environmental conditions and lighting features, and there are many types of cameras and ways to configure them. All these specific factors determine the success of deep learning.

Mats Thulin gave an example of what can be achieved by focusing on the specifics of a scenario: "Our traffic cameras can recognize license plates on vehicles with high accuracy. This kind of optimization will have a high impact on the work with analytics applications in the future."

He also mentioned a new Axis camera with advanced computing capabilities aimed at advanced analytics: “This opens up the possibility of creating hybrid analytics solutions, where some of the tasks are performed on the cameras themselves, which frees up server resources.”