Object Detection and Semantic Segmentation
This project is about automatically detecting and segmenting objects in RGB and RBGD images. While there has been tremendous progress in the recent years in solving these tasks, existing methods still do not generalize well to unseen environments and unseen objects other than those in the training set. This problem is called domain gap and its solution approaches are referred to as domain adaptation. In this introductory laboratory project, the student has to review the latest literature on semantic image segmentation, select a promising method and fine tune it to new object classes and formerly unseen environments. The topic can be continued towards more complex domain adaptation in follow-up projects.
Külső partner: Sörös Gábor, Nokia Bell Labs