Scientists from the Institute of Space and Information Technologies of the Siberian Federal University (SFU), together with colleagues from the Malaysian University of Utara, have developed a new video data storage model that significantly simplifies and speeds up the process of finding the necessary fragments in large volumes of video information. This innovation is especially important for processing data from surveillance cameras.
The new model, created under the leadership of Associate Professor Ekaterina Zhang, includes a comprehensive infrastructure for storing and managing video data. Its main advantage is the ability to quickly extract specific video fragments without having to view the entire volume of recorded information. This is achieved through automatic detection of frames containing specific objects or people, which are then stored along with metadata in a database.
Unlike traditional methods, where you had to watch the entire video stream to find the desired fragment, the new technology uses hashing, creating a "digital fingerprint" of each frame. This not only allows you to quickly find the necessary fragments, but also guarantees the immutability of the video recording, protecting it from fakes.
Read materials on the topic:
В России испытали квантовую технологию, способную шифровать данные невзламываемыми ключами
Росэлектроника представила миниатюрную камеру технического зрения
Отберет работу у аниматоров: нейросеть Kandinsky научили создавать видео на основе текста
