Teaching CVAIAC
Computer Vision and Artificial Intelligence for Autonomous Cars
Autumn 2025 · ETH Zurich · 6 ECTS
About the course
This course introduces the core computer vision techniques and algorithms that autonomous cars use to perceive the semantics and geometry of their driving environment, localize themselves in it, and predict its dynamic evolution. Emphasis is placed on techniques tailored for real-world settings, such as multi-modal fusion, domain-adaptive and outlier-aware architectures, and multi-agent methods.
- Lecturer
- Christos Sakaridis
- When
- Fridays 14:15–17:00
- Where
- HG D 5.2
Slides, exercise sheets, recordings, and exam materials are password-protected. Access credentials are announced in the lecture and on the course forum.
Lecture team
Lecturer
Teaching Assistants
Lectures
| Date | Time | Room | Topic | Slides | Video |
|---|---|---|---|---|---|
| 19.09.2025 | 14:15–17:00 | HG D 5.2 | Fundamentals of Autonomous Cars | Video | |
| 26.09.2025 | 14:15–17:00 | HG D 5.2 | Fundamental Computer Vision Architectures and Algorithms for Autonomous Cars | Video | |
| 03.10.2025 | 14:15–17:00 | HG D 5.2 | Fundamental Computer Vision Architectures and Algorithms (continued) | Video | |
| 10.10.2025 | 14:15–17:00 | HG D 5.2 | Semantic Segmentation | Video | |
| 17.10.2025 | 14:15–17:00 | HG D 5.2 | Depth Estimation | Video | |
| 24.10.2025 | 14:15–17:00 | HG D 5.2 | Object Detection | Video | |
| 31.10.2025 | 14:15–17:00 | HG D 5.2 | Instance Segmentation and Panoptic Segmentation | Video | |
| 07.11.2025 | 14:15–17:00 | HG D 5.2 | Unimodal 3D Object Detection | Video | |
| 14.11.2025 | — | — | No lecture — CVPR conference deadline | — | — |
| 21.11.2025 | 14:15–17:00 | HG D 5.2 | 3D Reconstruction and Localization | Video | |
| 28.11.2025 | 14:15–17:00 | HG D 5.2 | Domain Adaptation | Video | |
| 05.12.2025 | 14:15–17:00 | HG D 5.2 | Multi-modal 2D and 3D Object Detection | Video | |
| 12.12.2025 | 14:15–17:00 | HG D 5.2 | Visual Grounding, Anomaly Segmentation and Vehicle-to-Vehicle Communication | Video | |
| 19.12.2025 | 14:15–17:00 | HG D 5.2 | Multiple Object Tracking and Motion Prediction | Video |
Practical sessions
| Date | Time | Room | Topic | Slides | Video |
|---|---|---|---|---|---|
| 03.10.2025 | 10:15–12:00 | Online | Getting Started with Python and SLURM | Video | |
| 10.10.2025 | 10:15–12:00 | Online | Project 1 Introduction: Semantic Segmentation and Depth Estimation | Video | |
| 17.10.2025 | 9:15–11:00 | Online | Project 1: Attention Mechanisms and Transformers | Video | |
| 24.10.2025 | 10:15–12:00 | Online | Project 1 Q&A | — | Video |
| 31.10.2025 | 10:15–12:00 | Online | Project 1 Q&A | Video | |
| 07.11.2025 | 10:15–12:00 | Online | Project 1 Q&A | Video | |
| 14.11.2025 | 10:15–12:00 | Online | Project 2 Introduction: 3D Detection from Point Clouds | Video | |
| 21.11.2025 | 10:15–12:00 | Online | Project 2 Q&A | Video | |
| 28.11.2025 | 10:15–12:00 | Online | Project 2 Q&A | Video | |
| 05.12.2025 | 10:15–12:00 | Online | Project 2 Q&A | Video | |
| 12.12.2025 | 10:15–12:00 | Online | Project 2 Q&A | Video | |
| 19.12.2025 | 10:15–12:00 | Online | Project 2 Q&A | — |
Projects
-
Semantic segmentation and depth estimation
- Starts
- 10.10.2025
- Due
- 03.11.2025 · 23:59
Implement and train multi-task models for dense visual perception on real-world driving data.
Handout (PDF) -
3D object detection using LiDARs
- Starts
- 07.11.2025
- Due
- 15.12.2025 · 23:59
Detect vehicles and other agents in LiDAR point clouds from a multi-modal driving dataset.
Handout (PDF)
Prerequisites
- Linear algebra, multivariate calculus, and probability theory.
- Basic computer vision and machine learning.
- Programming proficiency in Python, PyTorch, scikit-learn, and scikit-image.
Exam & grading
Exam
- Examiner
- Christos Sakaridis
- Format
- Written session examination
- Duration
- 120 minutes
- Language
- English
- Permitted
- One A4 sheet of paper and a simple non-programmable calculator.
- Offered only in the session following the end of the course.
- A mock exam (short version) is provided, with and without solutions.
- Mock exam solutions are discussed in the lecture on 12.12.2025.
Grading
Projects 50% · Exam 50%
- Projects are mandatory and conducted in groups.
- Failing the projects results in failing the final grade.
- Students who do not pass the projects must de-register from the session exam.
Learning objectives
- Understand the principles of visual sensors used in autonomous cars.
- Distinguish the dominant architectural paradigms in modern visual perception models.
- Categorize visual perception tasks and explain the algorithms that solve them.
- Critically analyze current research in computer vision for autonomous driving.
- Reproduce and adapt state-of-the-art methods for visual perception.
- Independently develop new visual perception models for autonomous cars.