Spacetime Vision: Towards Unsupervised Learning in the 4D World

Computers are starting to „see” the world around them, while fast approaching human intelligence. Systems based on current GPU technology and AI learn from large amounts of data to imitate human understanding and even exceed our skills on certain tasks. That makes many applications of AI possible in virtually all areas of human activity. Despite the recent success of deep learning and computer vision, important challenges still remain, especially at the higher level of learning and semantic understanding. In the Spacetime Vision project we aim to solve some of these challenging research problems, such as learning in the absence of human supervision, automatic description of visual data into language, making cameras smart in the 4D world and intelligent monitoring of sites and structures using drones. We bring together some of the best research teams in Romania and Norway and, based on our research experience and results published at top international forums, we will open doors for new artificial intelligence technologies.