Estimating the pose of objects from a single image has many applications, ranging from autonomous driving over manipulation to multi-robot SLAM. Based on some recent work, we would like to investigate a novel approach to this that could potentially be very powerful.
Estimating the pose of objects from a single image has many applications, ranging from autonomous driving over manipulation to multi-robot SLAM. Based on some recent work, we would like to investigate a novel approach to this that could potentially be very powerful.
Develop a method that, given a training set of images of a certain object, are able to reliably detect the pose of that object in previously unseen images. We will provide some guidance based on our experience, but you will also be able to bring in your creativity. The optimal outcome is a publication (CVPR/RSS).
Develop a method that, given a training set of images of a certain object, are able to reliably detect the pose of that object in previously unseen images. We will provide some guidance based on our experience, but you will also be able to bring in your creativity. The optimal outcome is a publication (CVPR/RSS).
Titus Cieslewski ( titus at ifi.uzh.ch ), APPLY VIA EMAIL, ATTACH CV AND TRANSCRIPT (also Bachelor)! Preferred skills: Linux, Python, some Computer Vision background, TensorFlow/PyTorch or equivalent. This project will be co-supervised by Elia Kaufmann.
Titus Cieslewski ( titus at ifi.uzh.ch ), APPLY VIA EMAIL, ATTACH CV AND TRANSCRIPT (also Bachelor)! Preferred skills: Linux, Python, some Computer Vision background, TensorFlow/PyTorch or equivalent. This project will be co-supervised by Elia Kaufmann.