Skip to content

hopessugar/weed_detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 

Repository files navigation

Soybean Weed Detection Model

Overview

This project involves the detection and classification of soybean field components using a lightweight deep learning model based on MobileNetV2. The model identifies four classes: Soybean, Soil, Broadleaf, and Weed, and groups them into two categories:

  • Category 1 (Non-Weeds): Soil, Soybean, Broadleaf
  • Category 2 (Weeds): Weed

The aim is to assist farmers in managing weeds effectively and improving crop yields through precise identification.


Key Features

  • Lightweight Model: Built using MobileNetV2, ensuring efficient performance and deployment on resource-constrained devices.
  • Multi-Class Classification: Accurately differentiates between field components and weeds.
  • Data Augmentation: Preprocessed dataset with techniques to improve model robustness against variations in field conditions.
  • Deployment-Ready: Optimized for real-time applications, such as drones or automated sprayers.

How It Works

  1. Input: Field images are fed into the model.
  2. Processing: The model classifies the input image into one of the four classes.
  3. Output: Results guide weed management decisions, such as precision spraying.

Technologies Used

  • Model Architecture: MobileNetV2
  • Frameworks: TensorFlow/Keras
  • Preprocessing Tools: OpenCV for image resizing and augmentation.

Applications

  • Precision Agriculture: Identify and target weeds for selective herbicide application.
  • Field Monitoring: Real-time weed and crop health analysis using drones or IoT devices.

Future Enhancements

  • Expanding the dataset to cover more weed and crop varieties.
  • Improving performance under extreme environmental conditions.
  • Integrating the model with geographic information systems (GIS).

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published