Shape classification technology of pollinated tomato flowers for robotic implementation.

Area of Science:

• Agricultural Engineering
• Computer Science
• Plant Science

Background:

• Greenhouse tomato cultivation commonly uses insect pollination, manual vibration, or plant growth regulators.
• Insect pollination is the preferred natural method, but advancements in technology can enhance efficiency.

Purpose of the Study:

• To introduce and evaluate a novel pollination method utilizing Artificial Intelligence (AI) for flower classification via drones or robots.
• To develop an AI system capable of recognizing and classifying tomato flowers ready for pollination, even amidst movement and vibration.

Main Methods:

• Development of an AI image classification system employing a machine learning convolutional neural network (CNN).
• Implementation of an image analysis algorithm focused on pollination flower shape for the AI classifier.
• Experimental validation in a tomato greenhouse setting to assess classification accuracy and impact on fruit development.

Main Results:

• The AI classifier achieved a significant accuracy rate, crucial for effective pollination.
• The AI-driven pollination method resulted in the formation of optimally shaped tomato fruits, confirmed by fruiting rate analysis.
• The system demonstrated adaptability to challenges like plant movement and drone-induced vibrations.

Conclusions:

• The AI-powered drone/robot pollination system is a viable advancement for greenhouse tomato cultivation.
• This technology shows promise for improving pollination efficiency and fruit quality in agriculture.
• The AI flower classification approach is adaptable for smart farming applications across various crop species.