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 Videos

Optimizing land use and land cover classification with deep learning on multi-resolution datasets.

Alisha Raut1,2, Sagar Tomar3,4

  • 1CSIR-Central Electronics Engineering Research Institute, Pilani, 333031, India. alisha.ceeri23a@acsir.res.in.

Environmental Monitoring and Assessment
|May 2, 2026
PubMed
Summary

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

Cerebral rheumatoid vasculitis Masquerading as an intracranial neoplasm.

Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia·2026
Same author

Approach-based techniques of CT-guided percutaneous vertebral biopsy.

Diagnostic and interventional radiology (Ankara, Turkey)·2020
Same author

Multiple intrahepatic pancreatic pseudocyst (MIHPPs): an overlooked and misdiagnosed entity.

Gastroenterology and hepatology from bed to bench·2019
See all related articles

MobileNetV3, ResNet34, and GoogleNet deep learning models show high accuracy for land use and land cover (LULC) classification. Optimized training pipelines significantly improve performance for remote sensing applications.

Area of Science:

  • Remote Sensing
  • Computer Vision
  • Machine Learning

Background:

  • Accurate land use and land cover (LULC) classification is crucial for environmental monitoring, urban planning, and resource management.
  • Deep learning models offer advanced capabilities for analyzing complex geospatial data.
  • Evaluating state-of-the-art architectures is essential for optimizing LULC classification tasks.

Purpose of the Study:

  • To assess the performance of MobileNetV3, ResNet34, and GoogleNet for LULC classification.
  • To investigate the impact of transfer learning, data augmentation, and adaptive learning rate scheduling on model accuracy.
  • To compare model efficiency and accuracy across diverse remote sensing datasets.

Main Methods:

  • Three deep learning architectures (MobileNetV3, ResNet34, GoogleNet) were employed.
Keywords:
CNNDeep learningGoogleNetLand use and land coverMobileNetV3Remote sensingResNet34Transfer learning

Related Experiment Videos

  • Models were enhanced using transfer learning, data augmentation, and adaptive learning rate scheduling.
  • Performance was evaluated on the EuroSAT (Sentinel-2 imagery, 10 classes) and PatternNet (high-resolution aerial imagery, 38 classes) datasets.

    • Main Results:

    • MobileNetV3 achieved the highest accuracy (97.83% on EuroSAT, 99.23% on PatternNet) with fast inference.
    • ResNet34 demonstrated strong performance (97.56%, 99.06%) particularly for complex classes.
    • GoogleNet provided balanced performance (97.36%, 99.58%) across datasets.
    • An ablation study revealed that data augmentation, transfer learning, and learning rate scheduling boosted accuracy by 5-13%.

    Conclusions:

    • Optimized deep learning models, particularly MobileNetV3, are highly effective for LULC classification.
    • Enhancements like transfer learning and data augmentation significantly improve classification accuracy.
    • This research provides a foundation for advanced cross-domain remote sensing applications.