An evidential combination method with multi-color spaces for remote sensing image scene classification

Huang, Linqing; Zhao, Wangbo; Liew, Alan Wee-Chung; You, Yang

doi:10.1016/j.inffus.2022.12.025

Cited by 10 publications

(6 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The accurate recognition of landslide images is very thorny primarily due to their irregular shape patterns, but the shape information can be reflected on multi-scale visual saliency maps [27]. The "evidence theory involvement" challenge also indicates a diversified evidence set is vital to lead in the evidence theory [50]. Therefore, we develop a solution with channel-wise fusion, which is also the first usage of geological visual saliency for evidence set generation.…”

Section: A the Proposed Strategy To Involve Evidence Theorymentioning

confidence: 99%

“…The paper [86] remote sensing scene image classification framework, HHTL, is demonstrated in [87]. And [50] proposes ECMS, a multicolor spaces-based remote sensing image classification with an optimizable BPA function discounting weights. This paper choose their best performed network, the GSANet [88] in comparison study.…”

Section: Comparison Algorithmsmentioning

confidence: 99%

“…In experiments, four metrics (Overall Accuracy, Precision, True Positive Rate, and F-measure score) are involved to guarantee a fair comparison [50]:…”

Section: Evaluation Criteriamentioning

confidence: 99%

“…where T P , T N , F P , F N represent true positive, true negative, false positive and false negative. ViT-32 [82] 0.9528 0.9324 0.9606 0.9324 0.8961 0.9750 0.8961 0.9139 0.9677 0.9139 SCCovNet-AlexNet [83] 0.9531 0.9388 0.9607 0.9465 0.8961 0.9775 0.9356 0.9169 0.9690 0.9408 SCCovNet-VGG16 [83] 0.9549 0.9324 0.9606 0.9497 0.8961 0.9790 0.9368 0.9319 0.9677 0.9430 VGG-VD-16 [84] 0.9516 0.9296 0.9582 0.9320 0.8571 0.9750 0.8896 0.8919 0.9665 0.9103 MF 2 CNet [85] 0.9350 0.8688 0.9619 0.8688 0.9026 0.9475 0.9025 0.8854 0.9547 0.8853 EMTCAL [86] 0.9747 0.9430 0.9874 0.9430 0.9675 0.9775 0.9675 0.9551 0.9824 0.9551 HHTL [87] 0.9819 0.9737 0.9851 0.9736 0.9610 0.9900 0.9610 0.9673 0.9875 0.9673 ECMS-GSANet [50] ✓ 0.9910 0.9745 0.9975 0.9745 0.9935 0.9900 0.9935 0.9839 0.9937 0.9839 Proposed (RepVGG-B0 backbone) ✓ 0.9440 0.8182 0.9600 0.8967 0.9007 0.9320 0.9025 0.8575 0.9458 0.8966 Proposed (SwinTransV2-T backbone) ✓ 0.9964 0.9935 0.9975 0.9935 0.9935 0.9975 0.9935 0.9935 0.9975 0.9935 E. Comparative analyses 1) Efficiency analysis on network structures: One of this study's goals is to demonstrate the efficiency of evidence theory-coupled deep learning in remote sensing landslide image classification. Thus to begin with, this paper compares solely using a single image classification backbone network with the proposal employing the evidence theory (both channel-wise fusion and interpretability improved threebranched fusion are included).…”

Section: Evaluation Criteriamentioning

confidence: 99%

“…Nevertheless, at least two modals or datasets are required in the two mentioned methods, limiting their real-world applications. In [50], the evidence set is obtained via multi-color spaces, but their simple color space conversion is adverse to network accuracy. [51] and [52]'s ideas may be efficient, but they are not designed for image classification.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Deep Evidential Remote Sensing Landslide Image Classification With a New Divergence, Multiscale Saliency and an Improved Three-Branched Fusion

Zhang,

Cui,

2024

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

View full text Add to dashboard Cite

Hitherto, image-level classification on remote sensing landslide images has been paid attention to, but the accuracy of traditional deep learning-based methods still have room for improvement. The evidence theory is found efficient to boost the accuracy of neural networks, however, the present study argues three challenges that hinder the lead-in of this theory in deep landslide image classification. Aiming at the three problems, this study makes three improvements. For the interpretability and decision-invariance losses of three previous divergences, we propose a Belief Jensen-Renyi divergence with properties proven. To couple the evidence theory with deep remote sensing landslide image classification, a channel-wise multi-scale visual saliency fusion is developed. We additionally find that the channel-wise fusion is capable to reduce false recognition of networks as compared with original RGB images. To avoid decision failures in evidence-theoretic fusion process, we design an interpretability improved three-branched fusion. Experiments on Bijie Landslide dataset corroborate the synergistic benefits of the three improvements, where the proposal is compared with state-ofthe-art image classification backbone networks, remote sensing image scene classifiers, evidence fusion algorithms and versatile evidence-theoretic deep learning classifiers. We also evaluated the new method with two sort of image degradation, as well as an actual scenario in Luding County, China whose data is publicly available. The source code is at https://github.com/defzhangaa/ deeplandslideDS.

show abstract

Section: A the Proposed Strategy To Involve Evidence Theorymentioning

confidence: 99%

Section: Comparison Algorithmsmentioning

confidence: 99%

“…In experiments, four metrics (Overall Accuracy, Precision, True Positive Rate, and F-measure score) are involved to guarantee a fair comparison [50]:…”

Section: Evaluation Criteriamentioning

confidence: 99%