Abstract— In recent years, breakthroughs in deep learning, computer vision, and machine learning techniques have potential to transform and modernize how the crops are grown, cared and even predict yield. One of the problems every farmer encounters is invasive weeds that can kill or hinder the growth of crops by stealing water, nutrients, and sunlight from the plants. Another problem farmers face is predicting yield of crops. This is important for farmers to try to allocate resources, while maintain a low cost and maximize profits. With recent advancement in computer vision, predicting yield can potentially be done with a cost effective method using state of the art algorithms. In this thesis, I will apply new methods to solve problems that farmers have been facing for hundreds of years. A methodology will be developed to collect data for weed detection along with a pipeline to process the images. The data will be used to train state of the art object detection models such as YOLO, Faster R-CNN, and SSD Mobile. In order to find an optimal model for real time detection of weeds, I will develop a data collection methodology for prediction of crop yield. This work presents a machine vision system for weed detection in vegetable crops using outdoor images, avoiding lighting and sharpness problems during acquisition step. This development will be a module for a weed removal mobile robot with camera for lighting controlled conditions. The purpose of this paper is to develop a useful algorithm to discriminate weed, using image filtering to extract color and area features, then, a process to label each object in the scene is implemented, finally, a classification based on area is proposed, including sensitivity, specificity, positive and negative predicted values in order to evaluate algorithm performance.
|
DOI:
10.17148/IJARCCE.2023.12586
