International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 05 | May 2024
www.irjet.net
p-ISSN: 2395-0072
Weed Detection Using Machine Learning Mrs. Reshma Shivraj Bhalke1, Dr. A. A. Dandvate2, 1Student , Department of Computer engineering, Dhole Patil College of engineering, Pune, Maharashtra, India 2Head of Department , Department of Computer engineering, Dhole Patil College of engineering, Pune
Maharashtra, India ---------------------------------------------------------------------***--------------------------------------------------------------------3. LITERATURE REVIEW Abstract - Due to the variable plant spacing in vegetable plantations, weed identification is more difficult in vegetable than weed identification in crops. There has been minimal research on weed identification in vegetable plantations so far. In the numerous focuses on traditional crop weed identification approaches have mostly focused on detecting weeds directly. Nevertheless, weed species vary greatly. In contrast, this research provides a new method that blends Machine learning and video/ image processing technologies. The first step was to use a trained center Net model to detect veggies and create bounding boxes around them. The remaining green objects that fell out of the boundary boxes were then labeled as weeds. As a result, the structure concentrates solely on detecting vegetables, avoiding the handling of numerous weed species.
XIAOJUN JIN 1 , JUN CHE2 , AND YONG CHEN1 [1]“ Weed Identification Using Deep Learning and Image Processing in Vegetable Plantation”, Due to the variable plant spacing in vegetable plantations as well as crop, weed identification is more difficult than weed identification in crops. There has been minimal research on weed identification in vegetable plantations thus far. Traditional crop weed identification approaches have mostly focused on detecting weeds directly through traditional methods; nevertheless, weed species vary greatly. In contrast, this research provides a new method that blends machine learning and image processing technologies. The first step was to use a trained Center Net model to detect veggies and create bounding boxes around them. The remaining green objects that fell out of the boundary boxes were then labeled as weeds. As a result, the structure concentrates solely on detecting vegetables and weed. Furthermore, by reducing the amount of the training image data set and the complexity of weed detection, this technique can improve the weed identification performance and accuracy. A colour index-based segmentation was used in image processing to extract weeds from the backdrop. Genetic Algorithms (GAs) were used to determine and assess the colour index used, which was based on Bayesian classification error. The trained Center Net model had a precision of 95.6 percent, a recall of 95.0 percent, and an F1 score of 0.953 during the field test.
Key Words: Deep Learning, Image Processing, Weed Detection, Machine Learning, YOLO 3
1.INTRODUCTION Modern agriculture is becoming more reliant on computer-based systems. Various technical advances have opened new possibilities to gather information and use it in agriculture as well as in other subjects. Agriculture may not have traditionally been the first to implement the latest discoveries in technology, however, precision agriculture with localization such as Global Positioning System (GPS) and other information technologies are becoming everyday tools for farmers. Automated machines are starting to take over tedious tasks formerly performed only by humans.
Pignatti S, Casa R.2 , Harfouche A.2 , Huang W. 3 , Palombo A. “ Maize Crop And Weeds Species Detection By Using Uav perpectral Data”,[2] In order to use precision agriculture techniques like patch spraying, it’s necessary to monitor and map weeds within agricultural crops. Both environmentally and economically, precision and targeted weed eradication would be beneficial. When high spatial and spectral resolution data (i.e., from UAV platforms) is available, VNIR hyper spectral data can be a strong tool for performing effective weed monitoring and identification. This study investigates the spectral differences between crops and weeds in order to assess the potential of UAV hyper spectral data to distinguish maize crops from weeds and different types of weeds. During the 2016 growing season in Italy, UAV and field hyper spectral data were collected in a few corn fields. The results demonstrated that leaf chlorophyll and carotenoid content, extracted using spectral indices or inverting PROSAIL, may be used to
In the new era Economic and ecological benefits are the driving forces to implement new methods into agriculture to increase production. Balancing efficient farming and preservation of nature has traditionally been difficult.
2. MOTIVATION The technique used for increasing result to identify and reuse weed affected area for more seeding to increase production. In the agricultural field weed had detected by its properties such as its Size, Shape, Spectral Reflectance, Texture features and gives the result crop/weed. In this document they have demonstrated weed detection by its Size features as well as video processing.
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 1195