Tal Shoshan
Papers
2
Total Citations
9
H-Index
2
About
Tal Shoshan’s research lies at the intersection of agricultural robotics and computer vision, with a focused mission to automate the precision thinning of Medjool date bunches—a labor-intensive task critical for fruit quality. Her work tackles the dual challenge of robust perception and realistic data generation in unstructured orchard environments. Her most cited paper, “Segmentation and motion parameter estimation for robotic Medjool-date thinning” (2021, 7 citations), proposes a method to segment individual date fruits and estimate their motion, enabling a robotic arm to perform selective thinning without damaging the bunch. Building on this, her 2023 paper introduces a novel approach to synthesize artificial Medjool date bunch images by modeling fruit bunches in 3D using structural decomposition and Bézier curves. This synthetic data pipeline generates annotated datasets that can train deep learning models for fruit detection and manipulation, overcoming the scarcity of labeled real-world agricultural images. Though early in her career, Shoshan’s contributions are already shaping the path toward fully autonomous date farming, demonstrating how 3D modeling and simulation can accelerate the development of robust, field-ready robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2