Master’s Thesis / Final-Year Project in Machine Learning and Computer Vision
Posted 14hrs ago
Employment Information
Report this job
Job expired or something wrong with this job?
Job Description
Abschlussarbeit zu Few-Shot Computer Vision für menios kamerabasierte Speisenerkennung in der Gemeinschaftsverpflegung. Entwicklung und Erprobung eines praxistauglichen Prototyps für Kantinen.
Responsibilities:
- Develop a specific research question concerning camera-based food recognition
- Research and compare current approaches in few-shot learning, metric learning, computer vision, and visual similarity search
- Develop and compare models such as DINOv2, SigLIP, and lightweight models
- Develop a method for recognizing new dishes from a small number of reference images without retraining the entire model
- Define suitable metrics and test scenarios, conduct experiments, and systematically evaluate the results
- Work with public food datasets and data from real-world cafeteria environments
- Investigate challenges such as lighting, portion sizes, tableware, and different locations
- Work with the development team to transfer findings into a functional prototype
- Potential areas of focus: open-set few-shot recognition, domain shift, visually similar dishes, multi-dish recognition, and using the menu as prior knowledge
Requirements:
- Currently pursuing a degree in Computer Science, Data Science, Mathematics, or a comparable field
- Solid foundations in machine learning, deep learning, and computer vision
- Experience with Python; ideally also with PyTorch
- Interest in current computer vision research
- Structured planning, documentation, and evaluation of experiments
- Initial experience with image classification, embeddings, transfer learning, few-shot learning, or object detection is a plus
- Application documents submitted as a single PDF, including a cover letter, CV, and certificates
Benefits:
- Flexible working hours and remote work
- Part-time or full-time: we will find the right working model for you
- Free lunch three times a week
- Open culture, short decision-making processes, and agile working methods
- Plenty of opportunity to shape your work and contribute ideas
- Modern office in Hanover or Plauen
- Fresh fruit, excellent coffee or tea, and a location-specific selection of beverages
- Regular team events and outings
- Bike leasing
- Corporate benefits
- Company pension scheme and financial contributions toward it
- Numerous development and training opportunities
- Attendance at international conferences
- Top-floor office with rooftop terrace
















