Computer Vision III: Detection, Segmentation, and Tracking

- Proposal-based object detection (Faster-RCNN)
- One-stage detectors (YOLO, SSD, RetinaNet)
- Point-based detection
- Instance segmentation (Mask-RCNN)
- Semantic segmentation
- Panoptic segmentation
- Video object segmentation (OSVOS)
- Visual object tracking
- Multiple object tracking
- Graph neural networks for object tracking
- 3D object tracking
- Trajectory prediction

Upon completion of this module, students will have acquired theoretical concepts behind the object detection, segmentation and tracking algorithms. They will be able to implement their own solution to solve these practical real-world problems with deep learning.

IN2346 Introduction to Deep Learning
MA0902 Analysis für Informatiker
MA0901 Lineare Algebra für Informatiker

Knowledge of Python and Pytorch is a must to complete the course practical assignments.

The lectures will provide extensive theoretical aspects of neural networks and in particular deep learning architectures for the problems of object detection, segmentation and tracking.
The practical sessions will allow the students to get familiar with state-of-the-art algorithms (implemented in PyTorch) and the tricks of trainings them for the aforementioned Computer Vision tasks.

Projector, blackboard, PC

Links will be provided during the lecture

The exam takes the form of a written test. Questions allow to assess acquaintance with concepts and algorithms of deep learning for object detection, segmentation (both object as well as semantic) and multiple object tracking. Students demonstrate the ability to design, train, and optimize neural network architectures to solve these basic computer vision problems.

- Written test of 90 minutes at the end of the course.
- The students will be awarded a bonus in case they successfully complete all practical assignments. Their progress and/or final results will be presented in the form of a poster or an oral presentation.