Related Experiment Video
Updated: Aug 2, 2025
06:19
Integration of Animal Behavioral Assessment and Convolutional Neural Network to Study Wasabi-Alcohol Taste-Smell Interaction
Published on: August 16, 2024
459
Weed Detection Using Deep Learning: A Systematic Literature Review.
Nafeesa Yousuf Murad1, Tariq Mahmood1, Abdur Rahim Mohammad Forkan2
1Big Data Analytics Laboratory, Department of Computer Science, School of Mathematics and Computer Science, Institute of Business Administration, Karachi 75270, Pakistan.
Sensors (Basel, Switzerland)
|April 13, 2023
Summary
Early weed detection using artificial intelligence (AI) significantly aids farmers. This review analyzes deep learning (DL) and machine learning (ML) techniques for identifying weeds in crops, highlighting their performance and common applications.
Area of Science:
- Agricultural Science
- Computer Science
- Data Science
Background:
- Weeds pose a significant threat to crop yields and agricultural economies, necessitating advanced detection methods.
- The rise of artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL), offers promising solutions for early weed detection.
Approach:
- This study presents a systematic literature review (SLR) of state-of-the-art DL techniques for weed detection.
- The review analyzed 52 application papers and 8 survey papers published since 2015, focusing on ML/DL algorithms and image processing techniques.
Key Points:
- The review identified 34 unique weed types, 16 image processing techniques, and 11 DL algorithms (including 19 CNN variants).
- Support Vector Machines (SVM) and Convolutional Neural Networks (CNNs) demonstrated high accuracy (up to 99%) in weed detection.
- RGB imagery captured by drones, robots, and cell phones were commonly used, with sugar beet frequently serving as the reference crop.
Conclusions:
- Deep learning techniques have rapidly advanced weed detection capabilities since 2015.
- The findings provide a comprehensive analysis of ML/DL performance for weed detection, serving as a valuable resource for future research.
