2023
DOI: 10.1109/tgrs.2023.3314586
|View full text |Cite
|
Sign up to set email alerts
|

LW-IRSTNet: Lightweight Infrared Small Target Segmentation Network and Application Deployment

Renke Kou,
Chunping Wang,
Ying Yu
et al.

Abstract: It is a challenging task to separate infrared (IR) small targets from complex backgrounds quickly and accurately. Many kinds of literature have designed various feature fusion modules to further extract IR small target features. Although these designs are slightly helpful to the improvement of IR small target detection accuracy, they will cause a significant increase in network params and FLOPs. To minimize the computational complexity of the network and achieve industrial implementation while ensuring accurac… Show more

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
8
0

Year Published

2023
2023
2024
2024

Publication Types

Select...
8
1
1

Relationship

0
10

Authors

Journals

citations
Cited by 30 publications
(8 citation statements)
references
References 65 publications
0
8
0
Order By: Relevance
“…To thoroughly evaluate the segmentation performance of the proposed LightCF-Net, we used five evaluation metrics: Intersection over Union (IoU), Dice Similarity Coefficient (DSC), Sensitivity (SE), Specificity (SP), and Accuracy (ACC). Then, we further compared the proposed LightCF-Net against ten advanced models in segmentation efficiency and accuracy, including five large-scale models (U-Net [ 13 ], U-Net++ [ 52 ], CE-Net [ 53 ], DilatedSegNet [ 54 ], and Polyp-PVT [ 34 ]) and five lightweight models (UNeXt [ 55 ], AttaNet [ 56 ], LW-IRSTNet [ 57 ], DCSAU-Net [ 58 ], and PolypSeg+ [ 17 ]). Specifically, for a relatively fair comparison, we adopted the same strategy as Wu et al [ 17 ], appropriately reducing the channel numbers of large-scale models to achieve a comparatively smaller size.…”
Section: Resultsmentioning
confidence: 99%
“…To thoroughly evaluate the segmentation performance of the proposed LightCF-Net, we used five evaluation metrics: Intersection over Union (IoU), Dice Similarity Coefficient (DSC), Sensitivity (SE), Specificity (SP), and Accuracy (ACC). Then, we further compared the proposed LightCF-Net against ten advanced models in segmentation efficiency and accuracy, including five large-scale models (U-Net [ 13 ], U-Net++ [ 52 ], CE-Net [ 53 ], DilatedSegNet [ 54 ], and Polyp-PVT [ 34 ]) and five lightweight models (UNeXt [ 55 ], AttaNet [ 56 ], LW-IRSTNet [ 57 ], DCSAU-Net [ 58 ], and PolypSeg+ [ 17 ]). Specifically, for a relatively fair comparison, we adopted the same strategy as Wu et al [ 17 ], appropriately reducing the channel numbers of large-scale models to achieve a comparatively smaller size.…”
Section: Resultsmentioning
confidence: 99%
“…Therefore, we have conducted complexity analysis experiments (i.e. Params, FLOPs, and FPS [47], [48].) to comprehensively evaluate the performance of LDCNet under the same hardware conditions described in Section III-C2.…”
Section: ) Complexity Analysismentioning
confidence: 99%
“…Infrared images, compared to visible light images, generally contain less useful information [29,30]. In the early stages, the field predominantly relied on model-based algorithms [31], which exhibited poor performance. In recent years, with the advancement of deep learning and the successive release of public infrared small target datasets, deep learning-based infrared small target detection algorithms have made significant progress.…”
Section: Infrared Small Target Detection Networkmentioning
confidence: 99%