In Russia, the Oculus automatic search system for prohibited content has been launched. This was reported Vedomosti representative of the Main Radio Frequency Center (FSUE GRCHTS, subordinated to Roskomnadzor). He assured that the system is already “performing the tasks assigned to it in full: it detects violations of the law in images and video materials.”
According to the representative of the GRFC, in December 2022 the system was tested, and in January its integration with other monitoring tools of Roskomnadzor began. No testing details are given. The main task of the system in the GRFC was the identification of violations of Russian legislation in images and videos.
“The system recognizes images and symbols, illegal scenes and actions, analyzes text in photo and video materials. Oculus automatically detects such offenses as extremist themes, calls for mass illegal events, suicide, pro-drug content, LGBT propaganda,” explained the representative of the GRFC.
Prior to the introduction of the system, the GRCHTs analyzed prohibited content “mostly manually,” the representative of the department specified. “On average, operators processed 106 images and 101 videos per day. Oculus, on the other hand, will analyze more than 200,000 images per day (about three seconds per image),” the representative of the GRFC added. The authorities hope that this will “increase the effectiveness of identifying signs of violations.”
Until 2025, it is planned to improve the system. In particular, “the possibility of adding new classes and types of violations, as well as the function of determining the postures of people and their actions,” said the representative of the GRFC. The need to use an automated system in the department was explained by the growing flow of prohibited materials on the Internet, including those related to the military operation in Ukraine.
As Kommersant reported, in the summer of 2022, the structure of Roskomnadzor allocated 57.7 million rubles. for the development of the Oculus system. It was assumed that the system would find prohibited content not only in text form, but also in “scenes, combinations of objects, compositions of images, faces, statics and movement dynamics.” Konstantin Bulanov, director of digital technologies at the GRCHTs, said that the search for prohibited content would be carried out thanks to neural networks based on deep machine learning.
Details – in the material “Kommersant” “Not walking around yes” Oculus “”.