Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

Simple Staining Technique01:24

Simple Staining Technique

OverviewStaining techniques in microscopy enhance the visualization of microorganisms by increasing contrast and allowing the differentiation of cellular structures. Simple staining is one of the fundamental methods used to observe the basic morphological characteristics of microorganisms, including their size, shape, and arrangement. This method relies on the application of a single dye to stain the entire cell, producing a clear contrast between the cell and the background.FixationFixation is...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Prognostics by generalists: large language models for lithium-ion batteries health forecasting.

Scientific reports·2026
Same author

Criminal emotion detection framework using convolutional neural network for public safety.

Scientific reports·2025
Same author

Enabling personalized smart tourism with location-based social networks.

PeerJ. Computer science·2024
Same author

Enhancing underwater target localization through proximity-driven recurrent neural networks.

Heliyon·2024
Same author

Underwater Wireless Sensor Networks Performance Comparison Utilizing Telnet and Superframe.

Sensors (Basel, Switzerland)·2023
Same author

Acoustic Wave Reflection in Water Affects Underwater Wireless Sensor Networks.

Sensors (Basel, Switzerland)·2023

Related Experiment Video

Updated: Jun 26, 2026

Creating Dynamic Images of Short-lived Dopamine Fluctuations with lp-ntPET: Dopamine Movies of Cigarette Smoking
14:21

Creating Dynamic Images of Short-lived Dopamine Fluctuations with lp-ntPET: Dopamine Movies of Cigarette Smoking

Published on: August 6, 2013

18.3K

A few-shot learning method for tobacco abnormality identification.

Hong Lin1,2, Zhenping Qiang1, Rita Tse2

  • 1College of Big Data and Intelligent Engineering, Southwest Forestry University, Kunming, China.

Frontiers in Plant Science
|April 29, 2024
PubMed
Summary

This study introduces a Feature Representation Enhancement Network (FREN) to improve few-shot learning (FSL) for identifying tobacco diseases. The FREN method enhances feature representation, significantly boosting accuracy in plant disease identification, especially for visually similar conditions.

Keywords:
cross-domainfeature representationfew-shot learninginstance-embeddingtask-adaptationtobacco disease identification

More Related Videos

Automated Measurement of Pulmonary Emphysema and Small Airway Remodeling in Cigarette Smoke-exposed Mice
10:37

Automated Measurement of Pulmonary Emphysema and Small Airway Remodeling in Cigarette Smoke-exposed Mice

Published on: January 16, 2015

13.3K
Imaging Spatial Reorganization of a MAPK Signaling Pathway Using the Tobacco Transient Expression System
08:54

Imaging Spatial Reorganization of a MAPK Signaling Pathway Using the Tobacco Transient Expression System

Published on: March 20, 2016

9.8K

Related Experiment Videos

Last Updated: Jun 26, 2026

Creating Dynamic Images of Short-lived Dopamine Fluctuations with lp-ntPET: Dopamine Movies of Cigarette Smoking
14:21

Creating Dynamic Images of Short-lived Dopamine Fluctuations with lp-ntPET: Dopamine Movies of Cigarette Smoking

Published on: August 6, 2013

18.3K
Automated Measurement of Pulmonary Emphysema and Small Airway Remodeling in Cigarette Smoke-exposed Mice
10:37

Automated Measurement of Pulmonary Emphysema and Small Airway Remodeling in Cigarette Smoke-exposed Mice

Published on: January 16, 2015

13.3K
Imaging Spatial Reorganization of a MAPK Signaling Pathway Using the Tobacco Transient Expression System
08:54

Imaging Spatial Reorganization of a MAPK Signaling Pathway Using the Tobacco Transient Expression System

Published on: March 20, 2016

9.8K

Area of Science:

  • Agricultural Science
  • Computer Vision
  • Machine Learning

Background:

  • Accurate disease identification in valuable crops like tobacco is crucial but hindered by data limitations for deep learning approaches.
  • Few-shot learning (FSL) offers a solution for data deficiency but struggles with weak feature representation, leading to poor generalization and cross-domain challenges.
  • Existing works rarely address disease identification for tobacco, a significant agricultural commodity.

Purpose of the Study:

  • To develop a novel method, the Feature Representation Enhancement Network (FREN), to overcome the weak feature representation issue in FSL for plant disease identification.
  • To enhance feature representation through instance embedding and task adaptation techniques.
  • To create and utilize a new dataset for tobacco leaf abnormality identification and validate the FREN's effectiveness.

Main Methods:

  • Proposed a Feature Representation Enhancement Network (FREN) incorporating instance embedding (global pooling, Gaussian-like calibration) and task adaptation (self-attention).
  • Developed a novel Tobacco Leaf Abnormality (TLA) dataset with 16 categories and 1,430 images.
  • Validated FREN on the benchmark PlantVillage dataset and the TLA dataset, introducing a subcategory division strategy for multi-symptom diseases.

Main Results:

  • Achieved 66.04% accuracy on 10 tomato categories in PlantVillage (5-way, 1-shot).
  • Reached 45.5% and 56.5% accuracy on the TLA dataset across two settings.
  • Improved accuracy to 60.7% (16-way, 1-shot) and 81.8% (16-way, 10-shot) on TLA using the multi-symptom solution, demonstrating enhanced feature representation and generalization.

Conclusions:

  • The FREN significantly improves few-shot plant disease identification performance by enhancing feature representation, particularly for categories with high visual similarity.
  • The proposed method demonstrates strong generalization ability, showing reduced sensitivity to domain shifts.
  • The developed TLA dataset and the multi-symptom disease solution contribute to advancing tobacco disease identification research.