Komplexpraktikum Computergraphik und Visualisierung (WS 2021/22)
SWS 0/0/4
Modules
INF-B-510 (Vertiefung) INF
INF-B-520 (Vertiefung zur Bachelor-Arbeit) INF
INF-B-530 (Vertiefung) MI
INF-B-540 (Vertiefung zur Bachelor-Arbeit)t MI
INF-VMI-8 (Angewandte Graphische Datenverarbeitung) MI
INF-MA-PR (Master-Praktikum) INF
INF-E-4 (Master-Praktikum) MI
INF-04-KP (Komplexpraktikum)
MINF-04-KP-FG1 (Komplexpraktikum Fachgebiet 1)
IST-05-KP (Komplexpraktikum)
CMS-VC-TEA (Visual Computing Team Project)
INF-VERT2 (Vertiefung Künstliche Intelligenz)
You can use the Komplexpraktikum also in INF-VERT7 with a corresponding notice to Prof. Gumhold.
Organisation
To register for one of the topics, please enroll in the OPAL course.
The first appointment / kick-off will take place in the last week of October, 25th to 29th. The concrete date and time will be agreed upon with all participants.
Topic "VR Video Segmentation Annotation Tool"
Advisor: Sebastian Vogt , Benjamin Russig , Dmitrij Schlesinger
Annotating segmentation masks is a very time-consuming and tedious job. Doing this for each frame of a video individually further increases the effort considerably. The idea of this complex lab is to annotate the video with scribbles in VR. These scribbles are then converted to complete semantic segmentation masks by machine learning. See a video with more details here.
Requirements
- advanced knowledge of computer graphics or computer vision
- coding experience with (C++ and graphics APIs) or (python and machine learning frameworks)
- willingness to learn new things as you go (see below)
Nice-to-have
- experience with the chairs cgv framework
- experience with VR development
- experience with client-server / backend development (python)
Topic "Collaborative Scanning"
Advisor: Tianfang Lin,
3D scanners often produce incomplete data sets, due to occlusion, unfavorable surface reflectance properties, or geometric restrictions in the scanner setup. As several 3D indoor scenes were already acquired from laser scanners by our group, these scenes are incomplete, therefore we plan to make up for the incompleteness by extra scans from RGBD cameras. See more details in a video.
Basic Task:
In this project, the 3D data is represented by point cloud. The VR devices will be HTC Vive Pro2 and develop the program by using cgv-framework(C++). There are 4 subtasks as below:
- Register the existing 3D indoor scene and the real-world scene(by seethrough function) in VR environment.
- Capture new scans from RGB-D cameras.
- Manipulate the scan into a precise position in the existing 3D point cloud.
- Merge the existing point cloud and partial scan.
Requirements
- advanced knowledge of computer graphics or computer vision
- coding experience with (C++ and graphics APIs)
- willingness to learn new things as you go (see below)
Nice-to-have
- experience with the chairs cgv-framework
- experience with VR development
- experience with RGBD cameras(Kinect Azure, Intel Realsense, Artec Spider)