SGT-BST: A graph neural network approach for multi-object tracking of solid-colored cattle in controlled pasture settings

Key Points

• Tracking solid-colored cattle significantly improved using a graph neural network with over 85% accuracy in controlled environments.
• The analysis focuses on enhancing multi-object tracking methods specific to solid-colored cattle for better farm management.
• Implementation involved utilizing computer vision techniques within controlled pasture conditions to ensure accuracy and effectiveness.
• These improvements may enable better livestock monitoring; further studies are needed in varied settings to validate results.