CCDN-DETR: A Detection Transformer Based on Constrained Contrast Denoising for Multi-Class Synthetic Aperture Radar Object Detection
Abstract
:1. Introduction
- (1).
- We propose the adoption of a cross-scale coder instead of the original multilayer transformer coder. This innovative approach expands the information scale, facilitating profound information modeling and fusion for the detection of SAR targets with significant scale differences.
- (2).
- The selection scheme for input DETR decoder queries is improved to guide the model to select features with high classification and IoU scores. Additionally, we designed a more potent decoder to further optimize the recognition efficiency.
- (3).
- To address the finite nature of SAR data and their high signal-to-noise ratio, we introduced constrained contrast denoising in the training phase. Real and noise labels were integrated into the decoder layer for contrast training. The real labels were confined within a controlled range, aiding the network to concentrate on clear boundary contours.
- (4).
- We present an end-to-end SAR object-detection solution called CCDN-DETR. This novel multiclass detection transformer is specifically tailored to the SAR domain. The effectiveness of DETR for SAR object detection was demonstrated through experiments conducted on two extensive multi-category datasets.
2. Related Studies
2.1. SAR Object Detection Based on CNN
2.2. Detection Transformer (DETR)
2.3. Multiclass SAR Datasets
3. Algorithm Framework
3.1. Overall Framework
3.2. Cross-Scale Encoder Design
3.3. Query Selection Mechanism Optimization
3.4. Denoising Decoder
3.4.1. Constrained Contrastive Denoising Training
3.4.2. Decoder Layers
4. Experiments and Results
4.1. Experimental Setup
4.1.1. Dataset
4.1.2. Training Settings
4.2. Evaluation Metrics
4.3. Experiment Results
4.3.1. Comparison on the Joint Dataset
4.3.2. Comparison on the MSAR Dataset
4.4. Ablation Experiments
4.4.1. Denoising Training Module Ablation Experiments
4.4.2. Impact of Query Initialization and Quantity
4.4.3. Decoder Layer Quantity Selection
4.4.4. Encoder Selection
4.4.5. Comparison of Inference Speed
4.5. Visualization
5. Discussion
6. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Li, J.; Xu, C.; Su, H.; Gao, L.; Wang, T. Deep learning for SAR ship detection: Past, present and future. Remote Sens. 2022, 14, 2712. [Google Scholar] [CrossRef]
- Ren, S.; He, K.; Girshick, R.; Sun, J. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. IEEE Trans. Pattern Anal. Mach. Intell. 2017, 39, 1137–1149. [Google Scholar] [CrossRef]
- Yang, G.; Lei, J.; Xie, W.; Fang, Z.; Li, Y.; Wang, J.; Zhang, X. Algorithm/hardware codesign for real-time on-satellite CNN-based ship detection in SAR imagery. IEEE Trans. Geosci. Remote Sens. 2022, 60, 5226018. [Google Scholar] [CrossRef]
- Ge, Z.; Liu, S.; Wang, F.; Li, Z.; Sun, J. Yolox: Exceeding yolo series in 2021. arXiv 2021, arXiv:2107.08430. [Google Scholar]
- Bochkovskiy, A.; Wang, C.Y.; Liao, H.Y.M. Yolov4: Optimal speed and accuracy of object detection. arXiv 2020, arXiv:2004.10934. [Google Scholar]
- Lin, T.Y.; Goyal, P.; Girshick, R.; He, K.; Dollár, P. Focal loss for dense object detection. In Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy, 22–29 October 2017; pp. 2980–2988. [Google Scholar]
- Liu, W.; Anguelov, D.; Erhan, D.; Szegedy, C.; Reed, S.; Fu, C.Y.; Berg, A.C. Ssd: Single shot multibox detector. In Proceedings of the Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, 11–14 October 2016; Proceedings, Part I 14. Springer International Publishing: Cham, Switzerland, 2016; pp. 21–37. [Google Scholar]
- Miao, T.; Zeng, H.; Yang, W.; Chu, B.; Zou, F.; Ren, W.; Chen, J. An improved lightweight RetinaNet for ship detection in SAR images. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2022, 15, 4667–4679. [Google Scholar] [CrossRef]
- Hou, Z.; Cui, Z.; Cao, Z.; Liu, N. An integrated method of ship detection and recognition in SAR images based on deep learning. In Proceedings of the Igarss 2020–2020 IEEE International Geoscience and Remote Sensing Symposium, Waikoloa, HI, USA, 26 September–2 October 2020; IEEE: Piscataway, NJ, USA, 2020; pp. 1225–1228. [Google Scholar]
- Tang, J.; Cheng, J.; Xiang, D.; Hu, C. Large-difference-scale target detection using a revised Bhattacharyya distance in SAR images. IEEE Geosci. Remote Sens. Lett. 2022, 19, 4506205. [Google Scholar] [CrossRef]
- Shi, P.; Sun, H.; Xin, Y.; He, Q.; Wang, X. SDNet: Image-based sonar detection network for multi-scale objects. IET Image Process. 2023, 17, 1208–1223. [Google Scholar] [CrossRef]
- Peng, H.; Tan, X. Improved YOLOX’s anchor-free SAR image ship target detection. IEEE Access 2022, 10, 70001–70015. [Google Scholar] [CrossRef]
- Feng, Y.; You, Y.; Tian, J.; Meng, G. OEGR-DETR: A Novel Detection Transformer Based on Orientation Enhancement and Group Relations for SAR Object Detection. Remote Sens. 2023, 16, 106. [Google Scholar] [CrossRef]
- Zhang, Y.; Jiang, S.; Cao, Y.; Xiao, J.; Li, C.; Zhou, X.; Yu, Z. Hardware-Aware Design of Speed-Up Algorithms for Synthetic Aperture Radar Ship Target Detection Networks. Remote Sens. 2023, 15, 4995. [Google Scholar] [CrossRef]
- Cui, J.; Jia, H.; Wang, H.; Xu, F. A fast threshold neural network for ship detection in large-scene SAR images. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2022, 15, 6016–6032. [Google Scholar] [CrossRef]
- Yang, X.; Zhang, J.; Chen, C.; Yang, D. An efficient and lightweight CNN model with soft quantification for ship detection in SAR images. IEEE Trans. Geosci. Remote Sens. 2022, 60, 5230713. [Google Scholar] [CrossRef]
- Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A.N.; Polosukhin, I. Attention is all you need. Adv. Neural Inf. Process. Syst. 2017, 30. [Google Scholar]
- Xia, R.; Chen, J.; Huang, Z.; Wan, H.; Wu, B.; Sun, L.; Xing, M. CRTransSar: A visual transformer based on contextual joint representation learning for SAR ship detection. Remote Sens. 2022, 14, 1488. [Google Scholar] [CrossRef]
- Liu, Z.; Lin, Y.; Cao, Y.; Hu, H.; Wei, Y.; Zhang, Z.; Guo, B. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Montreal, BC, Canada, 11–17 October 2021; pp. 10012–10022. [Google Scholar]
- Sun, Y.; Wang, W.; Zhang, Q.; Ni, H.; Zhang, X. Improved YOLOv5 with transformer for large scene military vehicle detection on SAR image. In Proceedings of the 2022 7th International Conference on Image Vision and Computing (ICIVC), Xi’an, China, 26–28 July 2022; IEEE: Piscataway, NJ, USA, 2022; pp. 87–93. [Google Scholar]
- Jocher, G.; Chaurasia, A.; Stoken, A.; Borovec, J.; Kwon, Y.; Michael, K.; Jain, M. ultralytics/yolov5: v7.0-yolov5 sota realtime instance segmentation. Zenodo 2022. [Google Scholar] [CrossRef]
- Zha, M.; Qian, W.; Yang, W.; Xu, Y. Multifeature transformation and fusion-based ship detection with small targets and complex backgrounds. IEEE Geosci. Remote Sens. Lett. 2022, 19, 4511405. [Google Scholar] [CrossRef]
- Luo, R.; Zhao, L.; He, Q.; Ji, K.; Kuang, G. Intelligent technology for aircraft detection and recognition through SAR imagery: Advancements and prospects. J. Radars, 2024, in press. [CrossRef]
- Carion, N.; Massa, F.; Synnaeve, G.; Usunier, N.; Kirillov, A.; Zagoruyko, S. End-to-end object detection with transformers. In Proceedings of the European Conference on Computer Vision, Glasgow, UK, 23–28 August 2020; Springer International Publishing: Cham, Switzerland, 2020; pp. 213–229. [Google Scholar]
- Zhang, H.; Li, F.; Liu, S.; Zhang, L.; Su, H.; Zhu, J.; Shum, H.Y. Dino: Detr with improved denoising anchor boxes for end-to-end object detection. arXiv 2022, arXiv:2203.03605. [Google Scholar]
- Li, J.; Qu, C.; Shao, J. Ship detection in SAR images based on an improved faster R-CNN. In Proceedings of the 2017 SAR in Big Data Era: Models, Methods and Applications (BIGSARDATA), Beijing, China, 13–14 November 2017; IEEE: Piscataway, NJ, USA, 2017; pp. 1–6. [Google Scholar]
- Wei, S.; Zeng, X.; Qu, Q.; Wang, M.; Su, H.; Shi, J. HRSID: A high-resolution SAR images dataset for ship detection and instance segmentation. IEEE Access 2020, 8, 120234–120254. [Google Scholar] [CrossRef]
- Wang, Z.; Kang, Y.; Zeng, X.; Wang, Y.; Zhang, T.; Sun, X. SAR-AIRcraft-1.0: High-resolution SAR aircraft detection and recognition dataset. J. Radars 2023, 12, 906–922. [Google Scholar] [CrossRef]
- Hu, J.; Shen, L.; Sun, G. Squeeze-and-excitation networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 23–28 June 2018; pp. 7132–7141. [Google Scholar]
- Xu, X.; Zhang, X.; Zhang, T.; Shi, J.; Wei, S.; Li, J. On-board ship detection in SAR images based on L-YOLO. In Proceedings of the 2022 IEEE Radar Conference (RadarConf22), New York, NY, USA, 21–25 March 2022; IEEE: Piscataway, NJ, USA, 2022; pp. 1–5. [Google Scholar]
- Shin, S.; Kim, Y.; Hwang, I.; Kim, J.; Kim, S. Coupling denoising to detection for SAR imagery. Appl. Sci. 2021, 11, 5569. [Google Scholar] [CrossRef]
- Xu, P.; Li, Q.; Zhang, B.; Wu, F.; Zhao, K.; Du, X.; Zhong, R. On-board real-time ship detection in HISEA-1 SAR images based on CFAR and lightweight deep learning. Remote Sens. 2021, 13, 1995. [Google Scholar] [CrossRef]
- Li, F.; Zhang, H.; Liu, S.; Guo, J.; Ni, L.M.; Zhang, L. Dn-detr: Accelerate detr training by introducing query denoising. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA, 18–24 June 2022; pp. 13619–13627. [Google Scholar]
- Ma, C.; Zhang, Y.; Guo, J.; Hu, Y.; Geng, X.; Li, F.; Ding, C. End-to-end method with transformer for 3-D detection of oil tank from single SAR image. IEEE Trans. Geosci. Remote Sens. 2021, 60, 5217619. [Google Scholar] [CrossRef]
- Chen, Y.; Xia, Z.; Liu, J.; Wu, C. TSDet: End-to-End Method with Transformer for SAR Ship Detection. In Proceedings of the 2022 International Joint Conference on Neural Networks (IJCNN), Padua, Italy, 18–23 July 2022; IEEE: Piscataway, NJ, USA, 2022; pp. 1–8. [Google Scholar]
- Huang, L.; Liu, B.; Li, B.; Guo, W.; Yu, W.; Zhang, Z.; Yu, W. OpenSARShip: A dataset dedicated to Sentinel-1 ship interpretation. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2017, 11, 195–208. [Google Scholar] [CrossRef]
- Lei, S.; Lu, D.; Qiu, X.; Ding, C. SRSDD-v1. 0: A high-resolution SAR rotation ship detection dataset. Remote Sens. 2021, 13, 5104. [Google Scholar] [CrossRef]
- Chen, J.; Lei, B.; Song, Q.; Ying, H.; Chen, D.Z.; Wu, J. A hierarchical graph network for 3d object detection on point clouds. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, WA, USA, 13–19 June 2020; pp. 392–401. [Google Scholar]
- Yao, Z.; Ai, J.; Li, B.; Zhang, C. Efficient detr: Improving end-to-end object detector with dense prior. arXiv 2021, arXiv:2104.01318. [Google Scholar]
- Lv, W.; Xu, S.; Zhao, Y.; Wang, G.; Wei, J.; Cui, C.; Liu, Y. Detrs beat yolos on real-time object detection. arXiv 2023, arXiv:2304.08069. [Google Scholar]
- Lin, T.Y.; Dollár, P.; Girshick, R.; He, K.; Hariharan, B.; Belongie, S. Feature pyramid networks for object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA, 21–26 July 2017; pp. 2117–2125. [Google Scholar]
- Zhu, X.; Su, W.; Lu, L.; Li, B.; Wang, X.; Dai, J. Deformable detr: Deformable transformers for end-to-end object detection. arXiv 2020, arXiv:2010.04159. [Google Scholar]
- Wang, Y.; Zhang, X.; Yang, T.; Sun, J. Anchor detr: Query design for transformer-based detector. In Proceedings of the AAAI Conference on Artificial Intelligence, Virtual, 22 February–1 March 2022; Volume 36, pp. 2567–2575. [Google Scholar]
- Liu, H.; Chen, Q.; Tan, Z.; Liu, J.J.; Wang, J.; Su, X.; Wang, J. Group Pose: A Simple Baseline for End-to-End Multi-person Pose Estimation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Paris, France, 2–6 October 2023; pp. 15029–15038. [Google Scholar]
- Xu, S.; Wang, X.; Lv, W.; Chang, Q.; Cui, C.; Deng, K.; Lai, B. PP-YOLOE: An evolved version of YOLO. arXiv 2022, arXiv:2203.16250. [Google Scholar]
- Lyu, C.; Zhang, W.; Huang, H.; Zhou, Y.; Wang, Y.; Liu, Y.; Chen, K. Rtmdet: An empirical study of designing real-time object detectors. arXiv 2022, arXiv:2212.07784. [Google Scholar]
- Zong, Z.; Song, G.; Liu, Y. Detrs with collaborative hybrid assignments training. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Paris, France, 2–6 October 2023; pp. 6748–6758. [Google Scholar]
- He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778. [Google Scholar]
- Han, L.; Ran, D.; Ye, W.; Wu, X. Asymmetric convolution-based neural network for SAR ship detection from scratch. In Proceedings of the 2020 9th International Conference on Computing and Pattern Recognition, Xiamen, China, 30 October–1 November 2020; pp. 90–95. [Google Scholar]
- Liu, C.; Zhu, W. An improved algorithm for ship detection in SAR images based on CNN. In Proceedings of the Twelfth International Conference on Graphics and Image Processing (ICGIP 2020), Xi’an, China, 13–15 November 2020; SPIE: Bellingham, DC, USA, 2021; Volume 11720, pp. 63–71. [Google Scholar]
- Sun, Z.; Dai, M.; Leng, X.; Lei, Y.; Xiong, B.; Ji, K.; Kuang, G. An anchor-free detection method for ship targets in high-resolution SAR images. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2021, 14, 7799–7816. [Google Scholar] [CrossRef]
Method | Inference | Backbone | Parameters (M) | Precision | Recall | mAP50 | mAP50-95 |
---|---|---|---|---|---|---|---|
YOLO V5 [21] | ultralytics | CspDarkNet | 25.1 | 0.849 | 0.827 | 0.893 | 0.606 |
YOLO V8 | ultralytics | CspDarkNet | 25.8 | 0.873 | 0.849 | 0.918 | 0.623 |
RetinaNet [6] | ICCV2017 | ResNet-FPN | 19.8 | 0.786 | 0.757 | 0.879 | 0.418 |
RTMDet [46] | MMDetection | CspNext | 24.7 | 0.871 | 0.850 | 0.926 | 0.606 |
Faster R-CNN [2] | NeurIPS2015 | ResNet-FPN | 28.3 | 0.744 | 0.739 | 0.879 | 0.435 |
Deformable DETR [42] | ICLR2021 | ResNet [48] | 23.5 | 0.835 | 0.809 | 0.867 | 0.521 |
DINO-DETR [25] | ICLR2023 | ResNet [48] | 24.6 | 0.837 | 0.808 | 0.913 | 0.582 |
RT-DETR [40] | PaddlePaddle | HGNetV2 | 31.9 | 0.864 | 0.848 | 0.911 | 0.619 |
CO-DETR [47] | ICCV2023 | ResNet | 30.1 | 0.870 | 0.837 | 0.905 | 0.622 |
CCDN-DETR (ours) | / | HGNetV2 | 23.6 | 0.891 | 0.856 | 0.919 | 0.635 |
Method | Inference | Backbone | Parameters (M) | Precision | Recall | mAP50 | mAP50-95 |
---|---|---|---|---|---|---|---|
YOLO V5 [21] | ultralytics | CspDarkNet | 25.1 | 0.834 | 0.812 | 0.881 | 0.612 |
YOLO V8 | ultralytics | CspDarkNet | 25.8 | 0.870 | 0.852 | 0.925 | 0.629 |
RetinaNet [6] | ICCV2017 | ResNet-FPN | 19.8 | 0.776 | 0.752 | 0.878 | 0.414 |
RTMDet [46] | MMDetection | CspNext | 24.7 | 0.871 | 0.849 | 0.925 | 0.605 |
Faster R-CNN [2] | NeurIPS2015 | ResNet-FPN | 28.3 | 0.743 | 0.738 | 0.878 | 0.431 |
Deformable DETR [42] | ICLR2021 | ResNet [48] | 23.5 | 0.830 | 0.808 | 0.865 | 0.517 |
DINO-DETR [25] | ICLR2023 | ResNet [48] | 24.6 | 0.836 | 0.806 | 0.912 | 0.580 |
RT-DETR [40] | PaddlePaddle | HGNetV2 | 31.9 | 0.856 | 0.825 | 0.904 | 0.611 |
CO-DETR [47] | ICCV2023 | ResNet | 30.1 | 0.846 | 0.828 | 0.900 | 0.608 |
CCDN-DETR (ours) | / | HGNetV2 | 23.6 | 0.887 | 0.874 | 0.922 | 0.631 |
Method | Inference | Backbone | Parameters (M) | Precision | Recall | mAP50 | mAP50-95 |
---|---|---|---|---|---|---|---|
YOLO V5 [21] | ultralytics | CspDarkNet | 25.1 | 0.835 | 0.783 | 0.839 | 0.571 |
YOLO V8 | ultralytics | CspDarkNet | 25.8 | 0.837 | 0.788 | 0.843 | 0.579 |
RetinaNet [6] | ICCV2017 | ResNet-FPN | 19.8 | 0.665 | 0.592 | 0.570 | 0.345 |
RTMDet [46] | MMDetection | CspNext | 24.7 | 0.835 | 0.780 | 0.835 | 0.564 |
Faster R-CNN [2] | NeurIPS2015 | ResNet-FPN | 28.3 | 0.741 | 0.656 | 0.708 | 0.435 |
Deformable DETR [42] | ICLR2021 | ResNet [48] | 23.5 | 0.756 | 0.643 | 0.682 | 0.437 |
DINO-DETR [25] | ICLR2023 | ResNet [48] | 24.6 | 0.835 | 0.712 | 0.808 | 0.493 |
RT-DETR [40] | PaddlePaddle | HGNetV2 | 31.9 | 0.838 | 0.772 | 0.832 | 0.559 |
CO-DETR [47] | ICCV2023 | ResNet | 30.1 | 0.831 | 0.745 | 0.811 | 0.527 |
CCDN-DETR (ours) | / | HGNetV2 | 23.6 | 0.834 | 0.788 | 0.837 | 0.581 |
Method | Inference | Backbone | Parameters (M) | Precision | Recall | mAP50 | mAP50-95 |
---|---|---|---|---|---|---|---|
YOLO V5 [21] | ultralytics | CspDarkNet | 25.1 | 0.819 | 0.772 | 0.829 | 0.573 |
YOLO V8 | ultralytics | CspDarkNet | 25.8 | 0.825 | 0.784 | 0.841 | 0.577 |
RetinaNet [6] | ICCV2017 | ResNet-FPN | 19.8 | 0.648 | 0.602 | 0.563 | 0.342 |
RTMDet [46] | MMDetection | CspNext | 24.7 | 0.825 | 0.789 | 0.832 | 0.568 |
Faster R-CNN [2] | NeurIPS2015 | ResNet-FPN | 28.3 | 0.742 | 0.665 | 0.702 | 0.427 |
Deformable DETR [42] | ICLR2021 | ResNet [48] | 23.5 | 0.760 | 0.655 | 0.679 | 0.441 |
DINO-DETR [25] | ICLR2023 | ResNet [48] | 24.6 | 0.830 | 0.718 | 0.805 | 0.511 |
RT-DETR [40] | PaddlePaddle | HGNetV2 | 31.9 | 0.837 | 0.769 | 0.832 | 0.559 |
CO-DETR [47] | ICCV2023 | ResNet | 30.1 | 0.820 | 0.757 | 0.804 | 0.529 |
CCDN-DETR (ours) | / | HGNetV2 | 23.6 | 0.845 | 0.775 | 0.829 | 0.584 |
Method | Precision | Recall | mAP50 | mAP50-95 |
---|---|---|---|---|
BaseLine | 0.839 | 0.818 | 0.871 | 0.582 |
BaseLine (+DN) | 0.865 | 0.819 | 0.894 | 0.594 |
BaseLine (+CDN) | 0.884 | 0.832 | 0.912 | 0.622 |
CCDN-DETR (ours) | 0.891 | 0.856 | 0.919 | 0.635 |
Query Selection | Precision | Recall | mAP50 | mAP50-95 |
---|---|---|---|---|
Top-K (K = 100) | 0.877 | 0.843 | 0.905 | 0.625 |
Top-K (K = 300) | 0.874 | 0.845 | 0.903 | 0.626 |
IOU Top-K (K = 100) | 0.891 | 0.856 | 0.919 | 0.635 |
IOU Top-K (K = 300) | 0.894 | 0.851 | 0.922 | 0.638 |
Number of Decoder Layers | Precision | Recall | mAP50 | mAP50-95 | Parameters (M) | Latency (ms) |
---|---|---|---|---|---|---|
1 | 0.833 | 0.818 | 0.794 | 0.600 | 17.8 | 27.4 |
3 | 0.878 | 0.846 | 0.901 | 0.621 | 20.1 | 29.9 |
6 | 0.891 | 0.856 | 0.919 | 0.635 | 23.6 | 34.3 |
Encoder Selection | Precision | Recall | mAP50 | mAP50-95 | Parameters (M) | Latency (ms) |
---|---|---|---|---|---|---|
Multi-layer encoder (3 layers) | 0.849 | 0.816 | 0.899 | 0.601 | 15.5 | 32.2 |
Multi-layer encoder (6 layers) | 0.875 | 0.840 | 0.907 | 0.611 | 17.9 | 40.1 |
Cross-scale encoder | 0.891 | 0.856 | 0.919 | 0.635 | 23.6 | 34.3 |
Method | Precision | Recall | mAP50 | mAP50-95 | Parameters (M) | Latency (ms) |
---|---|---|---|---|---|---|
Faster-RCNN | 0.744 | 0.739 | 0.879 | 0.435 | 28.3 | 17.5 |
YOLO V8 | 0.873 | 0.849 | 0.918 | 0.623 | 25.8 | 14.9 |
Deformable DETR | 0.835 | 0.809 | 0.867 | 0.521 | 23.5 | 32.4 |
DION-DETR | 0.837 | 0.808 | 0.913 | 0.582 | 24.6 | 39.9 |
CCDN-DETR | 0.891 | 0.856 | 0.919 | 0.635 | 23.6 | 34.3 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Zhang, L.; Zheng, J.; Li, C.; Xu, Z.; Yang, J.; Wei, Q.; Wu, X. CCDN-DETR: A Detection Transformer Based on Constrained Contrast Denoising for Multi-Class Synthetic Aperture Radar Object Detection. Sensors 2024, 24, 1793. https://doi.org/10.3390/s24061793
Zhang L, Zheng J, Li C, Xu Z, Yang J, Wei Q, Wu X. CCDN-DETR: A Detection Transformer Based on Constrained Contrast Denoising for Multi-Class Synthetic Aperture Radar Object Detection. Sensors. 2024; 24(6):1793. https://doi.org/10.3390/s24061793
Chicago/Turabian StyleZhang, Lei, Jiachun Zheng, Chaopeng Li, Zhiping Xu, Jiawen Yang, Qiuxin Wei, and Xinyi Wu. 2024. "CCDN-DETR: A Detection Transformer Based on Constrained Contrast Denoising for Multi-Class Synthetic Aperture Radar Object Detection" Sensors 24, no. 6: 1793. https://doi.org/10.3390/s24061793
APA StyleZhang, L., Zheng, J., Li, C., Xu, Z., Yang, J., Wei, Q., & Wu, X. (2024). CCDN-DETR: A Detection Transformer Based on Constrained Contrast Denoising for Multi-Class Synthetic Aperture Radar Object Detection. Sensors, 24(6), 1793. https://doi.org/10.3390/s24061793