Weeds and Crop Image Classification using Deep Learning Technique

IEEE BASE PAPER ABSTRACT:

Growth of weeds in the agriculture land decrease the yield of the farm land. The automatic methods are very useful to detect the weeds and crops. Deep learning methods are being employed in a wide range of applications in agriculture and farming. In order to effectively manage weeds and increase yields, automatic weed detection and classification can be very helpful. Precision farming has the ability to identify weeds within the crop. Weed absorbs the nutrients, present in it and affects the crop. A key aspect in weed management is to contribute for better yields. The aim of the paper is to use deep learning techniques to increases the accuracy of weed detection and decrease farmer’s workload. In this work, classification of soybean crops and weeds (broadleaf, soil, grass) is done by using customized CNN approach. The proposed weeds and crop classification produced reliable results in terms of classification accuracy.

PROJECT OUTPUT VIDEO:

ALGORITHM / MODEL USED:

ResNet50.

OUR PROPOSED PROJECT ABSTRACT:

Weeds and Crop Image Classification using Deep Learning Technique is a significant project developed using MATLAB and employs the powerful ResNet50 algorithm/model. The primary objective of this project is to provide an efficient and automated solution for classifying images of crops and weeds in agricultural fields.

In modern agriculture, the ability to accurately differentiate between crops and weeds is critical for effective weed management, as it enables targeted and precise application of herbicides, reducing costs and minimizing the environmental impact. Manual identification and removal of weeds can be labor-intensive and time-consuming. Therefore, the utilization of deep learning techniques for image classification offers a promising solution to this problem.

This project leverages the ResNet50 model, a state-of-the-art convolutional neural network (CNN) architecture, renowned for its deep layers and exceptional performance in image recognition tasks. The ResNet50 model, pre-trained on a few dataset, is fine-tuned using a dataset containing images of crops and weeds. This fine-tuning process allows the model to adapt to the specific characteristics of crop and weed images, resulting in superior classification accuracy.

The achieved accuracy of 99.12% is a testament to the effectiveness of the proposed approach. This high accuracy level demonstrates the model’s capability to accurately distinguish between crops and weeds, minimizing false positives and negatives. The system’s performance surpasses many existing methods and showcases its potential for real-world deployment in agricultural settings.

This project holds great promise for revolutionizing weed management in agriculture, offering a cost-effective, time-saving, and environmentally friendly solution. The integration of deep learning and the ResNet50 model in MATLAB provides a robust platform for future advancements in agricultural automation, paving the way for sustainable and efficient farming practices.

SYSTEM REQUIREMENTS:

HARDWARE REQUIREMENTS:

System : Pentium i3 Processor.
Hard Disk : 500 GB.
Monitor : 15’’ LED.
Input Devices : Keyboard, Mouse.
Ram : 8 GB.

SOFTWARE REQUIREMENTS:

Operating system : Windows 10 Pro.
Coding Language : MATLAB
Tool : MATLABR2021A

REFERENCE:

Harshita Shri Panati; Gopika P; Diana Andrushia A; Mary Neebha T, “Weeds and Crop Image Classification using Deep Learning Technique”, 2023 9th International Conference on Advanced Computing and Communication Systems (ICACCS), IEEE Conference, 2023.

Matlab IEEE Projects

Weeds and Crop Image Classification using Deep Learning Technique