Working student / Master’s thesis: Zero-shot detection/segmentation from rendered templates

A common capability most of these robots require is the manipulation of objects. In this context, an important prerequisite is the detection and segmentation of these items. Yet, in ever-changing surroundings, robots are often encountered with newly, never-seen object instances. While there are often 3D models available, re-training a detector or segmentation network for every new instance is too cumbersome. As such, to be able to robustly identify such objects, new methods must be developed which can identify new instances on the fly based on a few rendered templates only, without the need of re-training.

We are looking for motivated, hands-on students to help our research group in the development and integration of a method on zero-shot detection and / or segmentation of object instances. This may be carried out as a working student position (up to 19hrs/month), or bachelor’s / master’s thesis.

Tasks:

• Identification of potential strategies of zero-shot detection / segmentation
• Implementation and training of the respective algorithm
• Integration and evaluation of the algorithm on a real-world robotic system

Qualifications:

• Knowledge about computer vision and / or machine learning
• Practical experience in the context of computer vision / machine learning in common libraries, e.g. OpenCV, Pytorch, Tensorflow
• Hands-on, dedicated, motivated attitude
• Basic knowledge about various programming / robotic software frameworks such as Python, ROS, git

What we offer:

• Motivated and growth-promoting environment
• Working with state-of-the-art algorithms
• Supervision by highly competent experts in the field
• Practical experience with real robotic systems
• Option for bachelor’s / master’s thesis