On-line real-time trunk detection, counting and sizing to enable precision agriculture tasks on a single-plant basis

To facilitate autonomous operations in orchards, an effective information management system is necessary. It should gather and process data on crop performance, including yield count, canopy volume, and crop health. A practical approach is to structure the system by ’discretizing’ it to individual trees, which requires tree segmentation/detection as a key component. This enables precise monitoring and analysis of each tree’s condition and productivity, aiding in informed decision-making and optimized orchard operations. The presented study wants to develop a low-cost approach to trunk detection, counting and sizing to possibly enable such an informed knowledgebase and decision making orchard management. The system relies on traditional computer vision algorithm to enable trunk segmentation by exploiting color and depth image information. Preliminary results are provided.