A multi-rotor UAV classification and recognition method based on RCSB-Net

To address the issue of low accuracy in target classification and recognition caused by the difficulty in extracting micro-Doppler time-frequency features from multi-rotor UAVs in complex low-altitude environments with strong clutter, this paper proposes a high-precision classification and recognition framework for multi-rotor UAVs, termed the RSPWVD, CBAM, Swin Transformer, and BiFPN based Network (RCSB-Net), which integrates RSPWVD-based time-frequency representation, CBAM-enhanced Swin Transformer feature extraction, and BiFPN-inspired bidirectional multi-scale feature fusion. This algorithm employs the Reassigned Smoothed Pseudo-Wigner–Ville Distribution (RSPWVD) to perform time-frequency processing on radar echo signals. Through an energy center-of-gravity reassignment mechanism, it suppresses cross-term interference and enhances time-frequency concentration, thereby preserving fine-grained micro-Doppler characteristics of the multi-rotor UAVs; Based on the Swin Transformer backbone network, a Convolutional Block Attention Module (CBAM) is embedded to enable the extraction of micro-Doppler features from targets in both the channel and spatial dimensions; a BiFPN-inspired bidirectional multi-scale feature fusion module is adopted to facilitate complementary interaction between shallow detailed features and deep semantic features ; A global feature aggregation and classification head is designed to predict target categories by combining multiscale features, thereby completing the classification of multi-rotor UAVs; Testing was conducted using the LSS-FMCWR-1.0 dataset. Experimental results show that RCSB-Net achieves a classification accuracy of 99.5%, a recall of 98.6%, and an F1-score of 91.3%. Compared with ResNet-50, the classification accuracy is improved by 6.1% points, demonstrating the effectiveness of the proposed framework for multi-rotor UAV classification.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-19
DOI
https://doi.org/10.1038/s41598-026-80874-7
Primary Topic
Advanced SAR Imaging Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A multi-rotor UAV classification and recognition method based on RCSB-Net

Jing Xing, Siqi Peng
Scientific Reports
Advanced SAR Imaging Techniques
article

A multi-rotor UAV classification and recognition method based on RCSB-Net

Jing Xing, Siqi Peng
article en

Abstract

To address the issue of low accuracy in target classification and recognition caused by the difficulty in extracting micro-Doppler time-frequency features from multi-rotor UAVs in complex low-altitude environments with strong clutter, this paper proposes a high-precision classification and recognition framework for multi-rotor UAVs, termed the RSPWVD, CBAM, Swin Transformer, and BiFPN based Network (RCSB-Net), which integrates RSPWVD-based time-frequency representation, CBAM-enhanced Swin Transformer feature extraction, and BiFPN-inspired bidirectional multi-scale feature fusion. This algorithm employs the Reassigned Smoothed Pseudo-Wigner–Ville Distribution (RSPWVD) to perform time-frequency processing on radar echo signals. Through an energy center-of-gravity reassignment mechanism, it suppresses cross-term interference and enhances time-frequency concentration, thereby preserving fine-grained micro-Doppler characteristics of the multi-rotor UAVs; Based on the Swin Transformer backbone network, a Convolutional Block Attention Module (CBAM) is embedded to enable the extraction of micro-Doppler features from targets in both the channel and spatial dimensions; a BiFPN-inspired bidirectional multi-scale feature fusion module is adopted to facilitate complementary interaction between shallow detailed features and deep semantic features ; A global feature aggregation and classification head is designed to predict target categories by combining multiscale features, thereby completing the classification of multi-rotor UAVs; Testing was conducted using the LSS-FMCWR-1.0 dataset. Experimental results show that RCSB-Net achieves a classification accuracy of 99.5%, a recall of 98.6%, and an F1-score of 91.3%. Compared with ResNet-50, the classification accuracy is improved by 6.1% points, demonstrating the effectiveness of the proposed framework for multi-rotor UAV classification.

Scientific Reports
Xi'an Peihua University (CN)
Advanced SAR Imaging Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon