Plant Instance Segmentation
2D plant segmentation
Deep learning based approach to recognize as well as segment individual plant instances from RGB-D images live or on-demand. In addition, there is also the possibility to consider other channels such as near-infrared. This approach can be executed live and produces a classification, bounding box and mask per detected plant, which can be used in further applications and devices.
Automatic 3D single plant segmentation.
Based on the automatic 3D plant mapping and 2D plant segmentation, plant instances can be extracted from the captured 3D plant map and reconstructed into 3D models. Based on this data, plant analysis can also be derived in the further process and actions based on it can be planned and executed for a specialized robotic platform.
- Creation of plant models for highly accurate analysis of various plant parameters, as well as for action planning and execution in the field or bed.
Automated segmentation of plants in images (2D) and based on point clouds (3D) allow plant-specific interaction with the environment (live or on-demand)