Interactive Segmentation is the problem of segmenting objects in an image with the help of user guidance. The goal is to create a segmentation mask which covers the area of a particular object in an image. In order to do so, the user provides some form of hint regarding the objects position to the model. Moste often these hints take the form of right / left mouse clicks that indicate that single pixels either belong to the background or foreground. Applications which make use of methods belonging to this problem are mostly related to data annotation or image editing software.

We focus on finding methods to aid interactive segmentation systems in performing on novel data.

For more information contact Robin Schön.

References:

• Robin Schön, Katja Ludwig and Rainer Lienhart. 2023. Impact of Pseudo Depth on Open World Object Segmentation with Minimal User Guidance. In 2023 IEEE/CVF International Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), June 17 2023 to June 24 2023, Vancouver, BC, Canada. IEEE, Piscataway, NJ, 4809-4819 CVPR Open Access 
• Robin Schön, Katja Ludwig, Julian Lorenz and Rainer Lienhart. 2024. Adapting the Segment Anything Model During Usage in Novel Situations. arXiv