HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR Images

Convolutional Neural Network (CNN)-based Synthetic Aperture Radar (SAR) target detection eliminates manual feature engineering and improves robustness but suffers from high computational costs, hindering on-satellite deployment. To address this, we propose HE-BiDet, an ultra-lightweight Binary Neura...

Full description

Saved in:
Bibliographic Details
Main Authors: Dezheng Zhang, Zehan Liang, Rui Cen, Zhihong Yan, Rui Wan, Dong Wang
Format: Article
Language:English
Published: MDPI AG 2025-04-01
Series:Micromachines
Subjects:
Online Access:https://www.mdpi.com/2072-666X/16/5/549
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850256467603488768
author Dezheng Zhang
Zehan Liang
Rui Cen
Zhihong Yan
Rui Wan
Dong Wang
author_facet Dezheng Zhang
Zehan Liang
Rui Cen
Zhihong Yan
Rui Wan
Dong Wang
author_sort Dezheng Zhang
collection DOAJ
description Convolutional Neural Network (CNN)-based Synthetic Aperture Radar (SAR) target detection eliminates manual feature engineering and improves robustness but suffers from high computational costs, hindering on-satellite deployment. To address this, we propose HE-BiDet, an ultra-lightweight Binary Neural Network (BNN) framework co-designed with hardware acceleration. First, we develop an ultra-lightweight SAR ship detection model. Second, we design a BNN accelerator leveraging four-directions of parallelism and an on-chip data buffer with optimized addressing to feed the computing array efficiently. To accelerate post-processing, we introduce a hardware-based threshold filter to eliminate redundant anchor boxes early and a dedicated Non-Maximum Suppression (NMS) unit. Evaluated on SAR-Ship, AirSAR-Ship 2.0, and SSDD, our model achieves 91.3%, 71.0%, and 92.7% accuracy, respectively. Implemented on a Xilinx Virtex-XC7VX690T FPGA, the system achieves <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>189.3</mn></mrow></semantics></math></inline-formula> FPS, demonstrating real-time capability for spaceborne deployment.
format Article
id doaj-art-c55d5f6bf0ae41b2867147f8ccacc547
institution OA Journals
issn 2072-666X
language English
publishDate 2025-04-01
publisher MDPI AG
record_format Article
series Micromachines
spelling doaj-art-c55d5f6bf0ae41b2867147f8ccacc5472025-08-20T01:56:38ZengMDPI AGMicromachines2072-666X2025-04-0116554910.3390/mi16050549HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR ImagesDezheng Zhang0Zehan Liang1Rui Cen2Zhihong Yan3Rui Wan4Dong Wang5Institute of Information Science, Beijing Jiaotong University, Beijing 100044, ChinaInstitute of Information Science, Beijing Jiaotong University, Beijing 100044, ChinaInstitute of Information Science, Beijing Jiaotong University, Beijing 100044, ChinaInstitute of Information Science, Beijing Jiaotong University, Beijing 100044, ChinaInstitute of Information Science, Beijing Jiaotong University, Beijing 100044, ChinaInstitute of Information Science, Beijing Jiaotong University, Beijing 100044, ChinaConvolutional Neural Network (CNN)-based Synthetic Aperture Radar (SAR) target detection eliminates manual feature engineering and improves robustness but suffers from high computational costs, hindering on-satellite deployment. To address this, we propose HE-BiDet, an ultra-lightweight Binary Neural Network (BNN) framework co-designed with hardware acceleration. First, we develop an ultra-lightweight SAR ship detection model. Second, we design a BNN accelerator leveraging four-directions of parallelism and an on-chip data buffer with optimized addressing to feed the computing array efficiently. To accelerate post-processing, we introduce a hardware-based threshold filter to eliminate redundant anchor boxes early and a dedicated Non-Maximum Suppression (NMS) unit. Evaluated on SAR-Ship, AirSAR-Ship 2.0, and SSDD, our model achieves 91.3%, 71.0%, and 92.7% accuracy, respectively. Implemented on a Xilinx Virtex-XC7VX690T FPGA, the system achieves <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>189.3</mn></mrow></semantics></math></inline-formula> FPS, demonstrating real-time capability for spaceborne deployment.https://www.mdpi.com/2072-666X/16/5/549binary neural networksship detectionsynthetic aperture radar (SAR)field programmable gate array (FPGA)
spellingShingle Dezheng Zhang
Zehan Liang
Rui Cen
Zhihong Yan
Rui Wan
Dong Wang
HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR Images
Micromachines
binary neural networks
ship detection
synthetic aperture radar (SAR)
field programmable gate array (FPGA)
title HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR Images
title_full HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR Images
title_fullStr HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR Images
title_full_unstemmed HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR Images
title_short HE-BiDet: A Hardware Efficient Binary Neural Network Accelerator for Object Detection in SAR Images
title_sort he bidet a hardware efficient binary neural network accelerator for object detection in sar images
topic binary neural networks
ship detection
synthetic aperture radar (SAR)
field programmable gate array (FPGA)
url https://www.mdpi.com/2072-666X/16/5/549
work_keys_str_mv AT dezhengzhang hebidetahardwareefficientbinaryneuralnetworkacceleratorforobjectdetectioninsarimages
AT zehanliang hebidetahardwareefficientbinaryneuralnetworkacceleratorforobjectdetectioninsarimages
AT ruicen hebidetahardwareefficientbinaryneuralnetworkacceleratorforobjectdetectioninsarimages
AT zhihongyan hebidetahardwareefficientbinaryneuralnetworkacceleratorforobjectdetectioninsarimages
AT ruiwan hebidetahardwareefficientbinaryneuralnetworkacceleratorforobjectdetectioninsarimages
AT dongwang hebidetahardwareefficientbinaryneuralnetworkacceleratorforobjectdetectioninsarimages