Understanding 3D environments is a long-term research topic for many years in computer vision. fundamental to understand 3D surroundings in many real-world computer vision applications, e.g. interactivity in robotics or AR/VR. With recent breakthroughs in deep learning, the computer vision community has made tremendous progress on perception in images. However, research in 3D perception has not been fully explored. Thanks to the commodity 3D sensors, such as the Kinect series, a variety of RGB-Datasets have been collected to enhance the 3D perception research. In this dissertation, we aim to investigate possible deep-learning-based solutions for 3D perception based on RGB-D data.
«
Understanding 3D environments is a long-term research topic for many years in computer vision. fundamental to understand 3D surroundings in many real-world computer vision applications, e.g. interactivity in robotics or AR/VR. With recent breakthroughs in deep learning, the computer vision community has made tremendous progress on perception in images. However, research in 3D perception has not been fully explored. Thanks to the commodity 3D sensors, such as the Kinect series, a variety of RGB-D...
»