Fruit maturity and location identification of beef tomato using R-CNN and binocular imaging technology

Abstract The objective of this study was to identify the maturity and position tomatoes in greenhouse. Three parts have been included study: building model image capturing object detection, identification mature fruits prediction size fruits. For first part, different time detection will be conducted greenhouse for second relative 3D calculated by binocular vision compared with actual measured position. third bounding box from fruit, correlation order pre-adjust width gripper plucking operation future. precision recall are over 95%. average error is 0.5 cm. R-squared 0.9.