New
Master Thesis - End-to-End 3D Reconstruction for Non-Pinhole Automotive Cameras (m/f/d)
Required skills
LinuxSLAMdeep learningneural networksPyTorchgeometric optimizationPythoncomputer visionMapAnythingcamera calibrationCOLMAPMASt3R3D reconstructionVGGTfisheye
Job description
Spleenlab posted this role. Below, we break down what it means for a working student in Saalburg-Ebersdorf: your weekly hours, take-home pay and visa limits. You can also open ChatGPT or Claude with a ready-made prompt to tailor your CV, check your fit, draft a cover letter or prep for the interview.
Will your CV clear this job's ATS filter? Scan it against this job with Jobscan
Description provided by Spleenlab
- Drones
We offer the solutions to equip your drone fleet with a new level of autonomy.
GPS-Denied Navigation
Surveillance | ISR
Collision Avoidance
Precision Landing
Vertical Clearance
Terrain-Following
Detect & Avoid
Drone Agent
- Vehicles
AI solutions for all types of robots — ground, humanoid, marine, trucks, and cars.
Multi-Object-Detection
Sensor Fusion
Freespace
Occupancy Grids
GPS-denied | SLAM
Collision Avoidance
Ground Agent
- Industries
ADAS & AD
Cargo | Logistics
Mining
Industrial
Defense ➚
Industries
Find the solution tailored to your industry, sector, or specific use case.
- Products
Auto Extrinsic Camera Calibration
Auto Multi Modal Sensor Calibration
Occupancy Grids
Products
AI solutions for all types of robots — drones, aircrafts, ground, humanoid, marine, trucks, and cars.
- Customers
We present success stories highlighting the impact of our software integrations in various customer projects.
Quantum Systems
Elmos Semiconductor SE
Rheinmetall
Krone | Lemken
Aerialoop
NXAI
Droneup
- Company
Stay updated with the latest company news, press releases, and upcoming events.
Careers
We’re looking for talent — explore our open job positions and join the team.
About
Discover more about Spleenlab, our global locations, and the work we do.
contact
Get in touch with us — we're here to help with any questions or project inquiries.
Events
Check out our event listings to see where you can meet us.
- Defense ➚
- GPS-Denied Navigation
Collision Avoidance
Precision Landing
Vertical Clearance
Terrain-Following
Detect & Avoid
Drone Agent
- Multi-Object-Detection
Freespace
Occupancy Grids
GPS-Denied | SLAM
Collision Avoidance
Ground Agent
- Agriculture
Cargo | Logistics
Mining
Industrial
Defense ➚
- VISIONAIRY® Auto Intrinsic Camera Calibration
VISIONAIRY® Auto Multi Modal Sensor Calibration
VISIONAIRY® Occupancy Grids
- VISIONAIRY® IR | EO Camera Depth Estimation
VISIONAIRY® Freespace | Collision Prevention
- VISIONAIRY® Single Object Tracking IR | EO
VISIONAIRY® Object Re-Identification
- VISIONAIRY® Follow Me Guidance
VISIONAIRY® Terminal Guidance
VISIONAIRY® Agents - Ground | Air | Sea
- VISIONAIRY® Visual Inertial Odometry
VISIONAIRY® SLAM - EO | IR
VISIONAIRY® SLAM - LiDAR
- SPLEENLAB® Video Stabilization
SPLEENLAB® AI Runtime
SPLEENLAB® Simulation
- Quantum Systems
Rheinmetall
Krone | Lemken
Aerialoop
NXAI
Droneup
- Newsroom
About
Contact
Events
Contact us
Defense ➚
Master Thesis - End-to-End 3D Reconstruction for Non-Pinhole Automotive Cameras (m/f/d)
Full-time
Saalburg-Ebersdorf | Jena
Software Engineering
Introductory information
This thesis evaluates modern end-to-end 3D reconstruction models for automotive multi-camera systems with non-pinhole camera models. Many learned reconstruction methods are designed around pinhole camera assumptions, while vehicle setups often use fisheye or double-sphere camera models.
The goal is to investigate strategies for adapting or refining such methods for robust camera-to-camera extrinsic calibration. Depending on the achieved results, this thesis may contribute to a scientific publication in computer vision, robotics, or autonomous systems.
Your Role
- Evaluate learned 3D reconstruction methods such as MASt3R, VGGT, or MapAnything.
- Analyze limitations caused by pinhole camera assumptions.
- Investigate adaptation strategies such as perspective crops, ray-based formulations, or post-hoc geometric refinement.
- Compare learned reconstruction methods against COLMAP-based calibration.
- Evaluate accuracy, robustness, scale consistency, and failure cases on vehicle data.
- Good programming skills in Python.
- Interest in deep learning, 3D reconstruction, and camera calibration.
- Familiarity with Linux-based development environments.
- Basic understanding of camera models or image-based 3D reconstruction.
- First experience with PyTorch or similar deep learning frameworks.
- Interest in fisheye, omnidirectional, or double-sphere camera models.
- Basic understanding of geometric computer vision.
- Motivation to combine learned methods with classical geometric optimization.
Are you ready to redefine the future of transportation across all domains? At Spleenlab, we're building the safety backbone for autonomous movement, ensuring people and things can navigate the world with confidence. Our expertise spans self-driving cars, trucks, maritime vessels, UAVs, air taxis, helicopters, and traditional aircraft.
We specialize in creating cutting-edge AI perception software powered by machine learning, specifically designed for safety-critical autonomous systems. This isn't just about code; it's about enabling a safer, more efficient tomorrow across land, air, and sea.
As part of our dynamic and interdisciplinary team, you'll work alongside recognized AI leaders, contribute to groundbreaking projects, and directly influence the next generation of autonomous mobility. If you're an ambitious AI professional eager to make a real difference, we want to hear from you.
Benefits
- Work-Life-Balance: Flexible working hours and mobile working options.
- Ergonomic Workstations: Equipped with height-adjustable desks and ergonomic office chairs.
- Company Events: Regular team events, meet & greets, and an annual Christmas party as integral parts of our company culture.
- Free Drinks: Enjoy complimentary coffee, tea, and water to keep you refreshed throughout the day.
At Spleenlab, we don’t just accept difference – we celebrate it, we support it, and we thrive on it for the benefit of our employees, our products and our community. We are committed to equal employment opportunity regardless of race, color, ancestry, religion, sex, national origin, sexual orientation, age, citizenship, marital status, disability, gender identity. We consider all qualified applicants, in accordance with the legal requirements. If you have a disability or special need that requires accommodation, please let us know.
Please send your CV and accompanying material directly to [email protected]
Apply now
View open positions
- For Drones
For Industries
- Product Overview
Newsletter
- News
Careers
Contact
Events
Specialized AI software company with a strong focus on safety-critical and defense-related applications.
Solutions
For Drones
For Vehicles
For Industries
Company
News
Careers
About
Contact
Events
Resources
Product Overview
VISIONAIRY® Suite
©
2026
Spleenlab GmbH | All Rights Reserved
Part of Quantum Systems Group
Privacy Policy
Legal Noctice
Terms of Service
Cookie Policy
- Defense ➚
Working student essentials
What this Tech working student role in Saalburg-Ebersdorf means for you: the weekly-hours rules, the social-contribution perks, and what international students should check before applying.
Weekly hours
Working students may work up to 20 hours a week during the semester and full-time during the breaks. Staying within this keeps your student status and the Werkstudent benefits.
Working student rulesSocial contributions
Under the Werkstudentenprivileg you're exempt from health, care and unemployment insurance contributions — only pension insurance applies. That leaves more net pay than a regular job.
Check your insuranceInternational students
Non-EU students can work 140 full or 280 half days per year (raised from 120/240 in March 2024). A working student contract usually fits within this — confirm the exact limits printed on your residence permit.
Studying in Germany
