Artificial Intelligence-Based Pollen Recognition and Counting System

Abstract

This artificial intelligence-based pollen recognition and counting system is designed to recognize and count healthy, unhealthy, and dead pollen in microscopic images with high accuracy and efficiency. This system uses a convolutional artificial neural network method that can detect objects in images quickly, precisely, and lightly. This system also has interactive software created with the Python programming language and the streamlit framework, which can display and store the results of pollen recognition and counting. This system is created by collecting, labeling, and augmenting pollen data, training and evaluating an artificial intelligence model with accuracy and error metrics, and integrating the model into the software. This system is useful for helping plant breeders optimize the crossing process.By utilizing this system, plant breeders can determine the quality and quantity of pollen more accurately and effectively, this is an effort to reduce the number of rejected seeds, which affects the productivity and sustainability of plantations.

Images