Improving the Bag-of-Words model with Spatial Pyramid matching using data augmentation for fine-grained arbitrary-oriented ship classification

Luu, Viet Hung; Dinh, Van Kiet; Hoa, Luong, Nguyen Hoang; Bui, Quynh; Nguyen, Tri-Thanh

doi:10.1080/2150704x.2019.1616123

Cited by 5 publications

(2 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, Luu et al [29] employed three data augmentation steps, including random rotation and flipping, to obtain additional remote sensing ship images for FGSC. Liu et al [12] proposed the local-aware CycleGAN to complete the image translation of the background and foreground, which can improve the reality of synthetic images.…”

Section: A Fgsc In Optical Remote Sensing Imagesmentioning

confidence: 99%

A Novel Multiscale Contrastive Learning Network for Fine-Grained Ocean Ship Classification

Dong,

Feng,

Fang

2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Fine-grained ocean ship classification plays a crucial role in maritime military surveillance, traffic management, and anti-smuggling operations. However, the complex backgrounds of remote sensing images (RSIs), as well as significant inter-class similarities and intra-class differences, result in poor classification performance. Hence, we propose MSCL-Net, a multi-scale contrastive learning network for fine-grained ship classification (FGSC). First, we introduce ResNet50 as the backbone network and extract the multi-layer features by using the FPN for FGSC. Second, a channel spatial attention module (CSAM) is proposed to extract the similarity (contrastive) feature of the same class, strengthening the representation learning ability for addressing issues caused by significant inter-class similarity and intra-class difference. Third, a region cropping and enlargement module (RCEM) is proposed to extract the fine-grained features of local discriminant regions in RSIs to overcome the challenge of background complexity. Finally, we used the CSAM to fuse the features of the original image and the cropped region image for FGSC. Additionally, we introduce a combined loss based on center loss and PolyLoss to enhance the discrimination ability of features and make it more suitable for the imbalance dataset compared with cross-entropy. We use a public finegrained ship classification dataset, FGSC-23, and our FGSC-41 to evaluate the performance of MSCL-Net. The experimental results show superior performance compared to other stateof-the-art methods, highlighting the effectiveness of MSCL-Net in addressing the challenges associated with fine-grained ship classification. Ablation experiments also suggest the effectiveness of our design.

show abstract

Section: A Fgsc In Optical Remote Sensing Imagesmentioning

confidence: 99%

A Novel Multiscale Contrastive Learning Network for Fine-Grained Ocean Ship Classification

Dong,

Feng,

Fang

2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Then, the principle of polar coordinate system is utilized for point location matching. Finally, the roads are decomposed into three types of primitive line elements and matched using curvature to improve the matching accuracy [5][6][7] . This method can realize the rapid real-time update of geographic information spatial database [8] .…”

Section: Introductionmentioning

confidence: 99%

Research on vector road data matching method based on flat curve

Yan,

Lu,

Sun

et al. 2024

Second International Conference on Environmental Remote Sensing and Geographic Information Technology (ERSGIT 2023)

View full text Add to dashboard Cite

In this research, a vector road data matching approach exclusively targeted at road data is proposed to suit the needs of quick and real-time updates of geographic databases. The location and kind of alterations to the road construction have to be precisely identified using this method. We suggest a technique for feature matching of vector road data from several eras under the same scale and geographic coordinate system by using the properties of road alignments. This helps us track changes in road development by allowing us to recognize how different road sections vary over time. Moreover, it forms the foundation for gradual updates of spatial data.

show abstract