Multi-feature Based Automated Flower Harvesting Techniques in Deep Convolutional Neural Networking

Abstract

In floriculture, automatic flower harvesting through robotic machines is in recent trends. Here in this research work we have proposed a calculation mechanism for automatic flower harvesting technique using deep convolutional neural network (DCNN). Here we also incorporate Faster R-CNN method which uses high-quality region proposals for flower detection from image generated by Region Proposal Network (RPN). Our aim is to make a fast, reliable and accurate system for detecting and harvesting the flower crop. The key concept of the system is to provide an automated robotic system which can detect the ripened flower crop and yield them; automatically without any human interference. Our research is based on the marigold flowers of different color and size. Here we described two methodologies first we present an approach for the flower detection from the image dataset captured from the camera and other for providing the accurate information to the automated robot to pluck the flowers from the plant and collect all them into the container.