| Topic: | Approximate camera pose estimation |
|---|
| Department: | Katedra kybernetiky |
|---|
| Supervisor: | Torsten Sattler |
|---|
| Announce as: | Diplomová práce, Bakalářská práce, Semestrální projekt |
|---|
| Description: | Visual localization is the problem of estimating the exact position and orientation from which an image was taken. Localization is a core component of applications such as self-driving cars and other autonomous robots, as well as Augmented Reality systems such as Microsoft HoloLens. The main part of a localization system is to determine the camera pose of a new image, i.e., the position and orientation from which the image was taken, with respect to a set of images with known pose. The goal of this project is to develop an approximate pose estimation algorithm based on simple relations such as visual overlap [1] that can be computed using deep neural networks. Datasets for training and testing are available. For more details, please contact the supervisor. |
|---|
| Bibliography: | [1] Rau et al., Predicting Visual Overlap of Images Through Interpretable Non-Metric Box Embeddings, ECCV 2020,
https://github.com/nianticlabs/image-box-overlap |
|---|
