基于机器视觉的大蒜鳞芽方向识别与自动调整装置设计OA

Aiming at the problems of difficult control of sprout orientation and low upright sprout rate in mechanized garlic sowing,a machine vision-based device for sprout orientation recognition and automatic adjustment was designed.With STM32F103 as the main controller,the ESP32-CAM vision module and a lightweight YOLO object detection algorithm were integrated.Image acquisition,annotation,and model training were completed through the Edge Impulse platform,and real-time recognition of sprout orientation was realized.The device was composed of a vision recognition module,a rotary adjustment mechanism,a servo drive,and a power supply system.Automatic discrimination of whether the sprout was facing left or right was performed,and the rotary groove was driven to complete directional adjustment,by which vertical upward sowing of the sprout was ensured.It was shown by the test results that an average recognition accuracy of over 95%for sprout orientation was achieved by the system,the posture adjustment was stable,and the upright sprout rate and operation efficiency of garlic sowing were significantly improved.The device was characterized by a compact structure and low cost,and technical support for the intelligent upgrading of garlic sowing machinery could be provided by it.