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 Experiment Video

Updated: Dec 4, 2025

Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
08:47

Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation

Published on: February 9, 2024

1.9K

PM2.5 concentration estimation using convolutional neural network and gradient boosting machine.

Zhenyu Luo1, Feifan Huang1, Huan Liu1

  • 1State Key Joint Laboratory of ESPC, School of the Environment, Tsinghua University, Beijing 100084, China; State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China.

Journal of Environmental Sciences (China)
|October 24, 2020
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

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

Sort by
Same author

Electronic Reconfiguration and Acidic Tailoring: Design of a CoMn Spinel Catalyst for Selective Catalytic Reduction-Coupled Chlorobenzene Deep Oxidation.

Environmental science & technology·2026
Same author

Study on the Influence and Mechanism of Resveratrol on Cognitive Impairment in Chronic Kidney Disease Rats Through Regulating Gut Microbiota and the TLR4/NFκB Pathway.

Journal of inflammation research·2025
Same author

Photon coupling-induced spectrum envelope modulation in the coupled resonators from Vernier effect to harmonic Vernier effect.

Nanophotonics (Berlin, Germany)·2024
Same author

Comment on 'Comparison of surgical outcomes and learning curve for robotic versus laparoscopic living donor hepatectomy: a retrospective cohort study'.

International journal of surgery (London, England)·2023
Same author

Generation of a human iPSC line (CIBi013-A) from a patient with young-onset Parkinson's disease carrying a novel homozygous PARK7 (DJ-1) mutation.

Stem cell research·2022
Same author

Effects of chronic kidney disease on cognitive function and α-klotho expression in hippocampus.

Translational andrology and urology·2022
Same journal

Interfacial engineering-mediated S-Scheme heterojunction with dual-ion cycling for enhanced photo-Fenton degradation of levofloxacin using a magnetically recyclable MnFe<sub>2</sub>O<sub>4</sub>@MIL-101(Fe) catalyst.

Journal of environmental sciences (China)·2026
Same journal

Corrigendum to "Quantifying carbon reduction potential of "Zero-Waste City" pilot: A case study of Shenzhen based on Source reduction-Recycling-Disposal framework" [Journal of Environmental Sciences, Volume 161, March 2026, Pages 411-420].

Journal of environmental sciences (China)·2026
Same journal

NO<sub>x</sub> regime-dependent effects of biogenic emissions on toluene degradation and product formation.

Journal of environmental sciences (China)·2026
Same journal

Greenhouse gas emissions from construction machinery in China: Historical trends and prospective reduction pathways.

Journal of environmental sciences (China)·2026
Same journal

Health risk of PM<sub>2.5</sub>-bound heavy metals in a megacity in South China: Comparison between before and after the outbreak of COVID-19.

Journal of environmental sciences (China)·2026
Same journal

Coupling nutrient limitation and light availability: Key pathways regulating phytoplankton primary productivity in urban lakes with different trophic statuses.

Journal of environmental sciences (China)·2026
See all related articles

This study introduces an image-based deep learning method for estimating fine particle (PM2.5) concentrations. The approach enhances air quality monitoring accessibility, especially in data-scarce regions.

Area of Science:

  • Environmental Science
  • Computer Science
  • Atmospheric Chemistry

Background:

  • Traditional PM2.5 monitoring relies on surface stations, vertical observations, and chemical transport models.
  • These methods have limitations in spatial coverage and accessibility, particularly in regions lacking ground-based infrastructure.

Purpose of the Study:

  • To develop and validate an image-based deep learning methodology for estimating PM2.5 concentrations.
  • To assess the transferability and accessibility of the proposed image-based approach for air quality monitoring.

Main Methods:

  • An end-to-end model combining Convolutional Neural Networks (CNN) and Gradient Boosting Machine (GBM) was trained using 6976 images, weather data, and hourly surface PM2.5 concentrations from Shanghai.
  • Transferability was tested by fine-tuning the model with 10% of image data from other locations.
Keywords:
Convolutional neural networkDeep learningHybrid modelPM(2.5) concentration

More Related Videos

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
08:27

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Published on: January 5, 2024

1.4K

Related Experiment Videos

Last Updated: Dec 4, 2025

Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
08:47

Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation

Published on: February 9, 2024

1.9K
Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
08:27

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Published on: January 5, 2024

1.4K
  • Feature importance analysis in GBM quantified image region sensitivity to PM2.5.

    • Main Results:

    • The model achieved high accuracy with a Mean Absolute Error (MAE) of 3.56, Root-Mean-Square Error (RMSE) of 10.02, and R-squared of 0.85.
    • Fine-tuning with minimal data (10%) demonstrated strong transferability to new locations.
    • The method identified image regions most indicative of PM2.5 levels.

    Conclusions:

    • The image-based deep learning approach offers a highly accessible and accurate method for PM2.5 estimation, overcoming limitations of traditional monitoring.
    • This methodology significantly improves the potential for widespread air quality assessment, especially in data-limited areas.
    • The study pioneers the use of graph theory and deep learning for innovative pollution monitoring.