Object detection and segmentations for automated sorting of specific objects on the conveyor belt, while achieving high generalization level with small number of data samples.
Automated object sorting on the assembly line with AI-based methods for object recognition and segmentation.
For manual labeling of image data a software tool was developed for faster data preparation. Transfer learning approaches and Mask R-CNN enable object detection with high confidence through efficient training and integration of new objects with small data set.
Real setup can be used as a testbed for various industrial applications.