Scholarlink

<p>In recent years, the Remote sensing (RS) scene classification leveraging deep learning (DL) is a fast-evolving domain with significant advancements. DL approaches have proven extremely effectual in extracting complex features and patterns from remote sensing images (RSI), enabling accurate and automated classification. The main advantage of DL for RS scene classification is its ability to learn a hierarchical representation from the raw information directly. Classical machine learning (ML) techniques often require manual feature engineering, where domain-specific features must be handcrafted. This manuscript presents a Sparrow Search Algorithm with Deep Learning Assisted Remote Sensing Scene Detection and Classification (SSADL-RSSDC). The proposed approach focuses on the automatic detection of numerous scene labels in RSI. Initially, the median filtering (MF) module is utilized for pre-processing. Furthermore, the SSADL-RSSDC system employs a deep residual network (ResNet) module to acquire the hierarchical module of the input image. Furthermore, the SSA has been implemented to choose the optimal values of Deep ResNet hyperparameters. At last, the extreme learning machine (ELM) approach can be utilized for the classification. The proposed method could be assessed under the UCM and AID database. The experimental assessment of the SSADL-RSSDC approach exhibited a greater accuracy value of 95.23% and 95.64% with dual database.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Leveraging sparrow search algorithm with deep learning assisted remote sensing scene detection and classification

  • 期刊论文

  • 发表日期：

    2026年6月15日

查看全文
  • M. Rega,
  • S. Sivakumar

摘要

In recent years, the Remote sensing (RS) scene classification leveraging deep learning (DL) is a fast-evolving domain with significant advancements. DL approaches have proven extremely effectual in extracting complex features and patterns from remote sensing images (RSI), enabling accurate and automated classification. The main advantage of DL for RS scene classification is its ability to learn a hierarchical representation from the raw information directly. Classical machine learning (ML) techniques often require manual feature engineering, where domain-specific features must be handcrafted. This manuscript presents a Sparrow Search Algorithm with Deep Learning Assisted Remote Sensing Scene Detection and Classification (SSADL-RSSDC). The proposed approach focuses on the automatic detection of numerous scene labels in RSI. Initially, the median filtering (MF) module is utilized for pre-processing. Furthermore, the SSADL-RSSDC system employs a deep residual network (ResNet) module to acquire the hierarchical module of the input image. Furthermore, the SSA has been implemented to choose the optimal values of Deep ResNet hyperparameters. At last, the extreme learning machine (ELM) approach can be utilized for the classification. The proposed method could be assessed under the UCM and AID database. The experimental assessment of the SSADL-RSSDC approach exhibited a greater accuracy value of 95.23% and 95.64% with dual database.