Deep learning for Root System Extraction from Barley Plants

The analysis of plant morphology, physiology, and behavior provides researchers with profound insights into how plants respond to diverse environmental conditions, pests, and diseases. This understanding is crucial for cultivating resilient crop varieties capable of enduring unfavorable circumstances, thereby bolstering sustainable agriculture and global food security. This study introduces an innovative method for extracting barley plant root systems from high-resolution images, representing a significant advancement in plant biology. The methodology encompasses several crucial stages, such as image segmentation, which plays a fundamental role in isolating roots from other image components, laying the groundwork for subsequent analysis. Subsequently, skeleton graph construction generates a topological representation of the root system, facilitating the interpretation of its intricate structure and connectivity patterns. Root path analysis further enhances understanding by tracking and assessing the dynamic characteristics of the root system within its environmental context. For segmentation, CafeNet was chosen over alternative methods due to its superior ability to generate accurate yet improvable masks. For root path analysis, the adoption of a customized approach for obtaining plant paths, rather than relying solely on the shortest path method, enables more precise values. Evaluation of the proposed approach against established methods such as RootNav 2.0 reveals notable enhancements in root system reconstruction accuracy across various performance metrics.