A Novel Temporal Dynamic Multi-Scale Spiking Transformer for Aerial Scene Classification
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Aerial scene understanding is important for applications involving land use analysis, environmental monitoring and urban planning based on highresolution drone and satellite imagery.Existing methodologies are limited due to the temporal variation, noise sensitivity, and the limited capability of regional generalization.To overcome these challenges, an EfficientNet-enhanced Temporal Dynamic Multi-scale Spiking Transformer (TDMST) framework is proposed to facilitate adaptive and energy-efficient classification of aerial scenes.EfficientNetV2 extracts low-and high-level features.The TDMST framework further processes the extracted features using multi-branch spiking neural mechanisms, along with transformer-like global attention mechanisms.Hyperparameters and architectural features are tuned using the Perfumer Optimization Algorithm (POA).Experiments conducted on the WHU-RS19, RSSCN7, and AID datasets achieved accuracies of 99.7%, 99.6%, and 99.3% respectively, demonstrating the TDMST's ability to securely and reliably classify aerial scenes, making it incredibly suitable for remote sensing applications.
Fields of Study & Subjects
Remote-Sensing Image Classification
Advanced Neural Network Applications
Remote Sensing and Land Use
Publication details
- Journal / source
- Advances in intelligent systems research/Advances in Intelligent Systems Research
- Pages
- 531–541
- Publication date
- 1 April 2026
- Language
- EN
- Document type
- Book chapter