RFA-Net: Reconstructed Feature Alignment Network for Domain Adaptation Object Detection in Remote Sensing Imagery

Zhu, Yangguang; Sun, Xian; Diao, Wenhui; Li, Hao; Fu, Kun

doi:10.1109/jstars.2022.3190699

Cited by 18 publications

(4 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Meanwhile, the key part detection task in military scenarios also faces typical challenges encountered in object detection in UAV images, such as domain shift, small objects, and limited real-time performance. Many highly inspiring studies [17][18][19] are proposed to alleviate these problems, which hold immense significance in fostering the progress of drone image object detection in various fields. For instance, D. Biswas et al [20] substantially reduced the number of learnable parameters of YOLOv5 through compressed convolutional techniques, transfer learning, and backbone shrinkage.…”

Section: A Key Part Identificationmentioning

confidence: 99%

A Classwise Vulnerable Part Detection Method for Military Targets

Wang,

Shen,

et al. 2024

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

View full text Add to dashboard Cite

Accurate vulnerable part detection based on full target detection results shows great importance in improving the damage effectiveness of the military drone. However, traditional object detection methods have difficulty in handling inaccurate full target bounding boxes and fail to model the semantic relationships between various class full targets and their key parts, resulting in low localization accuracy. The proposed approach includes a class-wise feature recalibration module, which effectively models the dependencies between the prior knowledge obtained from the full target detector and the location of the key part. Additionally, an optimized spatial transformation module is designed to preprocess the input image and eliminate interfering objects. Furthermore, a carefully constructed loss function is employed, linking the classification branch with the regression branch, thereby emphasizing the importance of localization accuracy. Our proposed model surpasses the performance of existing state-of-the-art models, demonstrating a significant advantage with maximum improvements of +24.9%, +30.2%, and +28.3% in mean Average Precision on the standard test set, generalized test set, and real-world dataset, respectively. The effectiveness and robustness are also confirmed through extensive ablation studies.

show abstract

Section: A Key Part Identificationmentioning

confidence: 99%

A Classwise Vulnerable Part Detection Method for Military Targets

Wang,

Shen,

et al. 2024

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

View full text Add to dashboard Cite

show abstract

“…Domain adaptation for object recognition in consumer image datasets successfully addresses weather, lighting conditions, geological variance, variation in image quality, and cross-camera adaptation by aligning the global feature distribution of data from the origin and target domains [15]. Recent State-of-the-art (SOTA) work of unsupervised domain adaptation for aerial imagery uses the reconstructed feature alignment method instead of adversarialbased feature alignment to avoid background noise alignment [16]. Nevertheless, limited progress in the domain adaptation task has been focused on satellite imagery.…”

Section: Domain Adaptation With Contrastive Learning For Object Detec...mentioning

confidence: 99%

Domain Adaptation with Contrastive Learning for Object Detection in Satellite Imagery

Biswas,

Tesic

2023

Preprint

View full text Add to dashboard Cite

<p>State-of-the-art object detection methods applied to satellite and drone imagery largely fail to identify small and dense objects. One reason is the high variability of content in the overhead imagery due to the terrestrial region captured and the high variability of acquisition conditions. Another reason is that the number and size of objects in aerial imagery are very different than in the consumer data. In this work, we propose a small object detection pipeline that improves the feature extraction process by spatial pyramid pooling, cross-stage partial networks, heatmap-based region proposal network, and object localization and identification through a novel image difficulty score that adapts the overall focal loss measure based on the image difficulty. Next, we propose novel contrastive learning with progressive domain adaptation to produce domain-invariant features across aerial datasets using local and global components. We show we can alleviate the degradation of object identification in previously unseen datasets. We create a first-ever domain adaptation benchmark using contrastive learning for the object detection task in highly imbalanced satellite datasets with significant domain gaps and dominant small objects from existing satellite benchmarks—the proposed method results in up to a 7.4% increase in mAP performance measure over the best state-of-art. </p>

show abstract

“…To address these problems, developing an algorithm that can transfer the knowledge learned from a labeled source domain to another unlabeled target domain becomes critical for object detection (Yao et al 2021;Li et al 2022c). Therefore, Unsupervised Domain Adaptation (UDA) is applied to the field of object detection (Zhang et al 2021a;Liu et al 2021a;Jin et al 2021;Zhu et al 2022) and a novel task called Domain Adaptive Object Detection (DAOD) (Chen et al 2018) is proposed to narrow the domain shift between two domains. Existing methods (Vs et al 2021;Zhao and Wang 2022;Xu et al 2022a) use Gradient Reversal Layers (GRL) (Ganin and Lempitsky 2015) and domain discriminators for this purpose.…”

Section: Introductionmentioning

confidence: 99%

Triple Feature Disentanglement for One-Stage Adaptive Object Detection

Wang,

Jia,

Zeng

et al. 2024

AAAI

View full text Add to dashboard Cite

In recent advancements concerning Domain Adaptive Object Detection (DAOD), unsupervised domain adaptation techniques have proven instrumental. These methods enable enhanced detection capabilities within unlabeled target domains by mitigating distribution differences between source and target domains. A subset of DAOD methods employs disentangled learning to segregate Domain-Specific Representations (DSR) and Domain-Invariant Representations (DIR), with ultimate predictions relying on the latter. Current practices in disentanglement, however, often lead to DIR containing residual domain-specific information. To address this, we introduce the Multi-level Disentanglement Module (MDM) that progressively disentangles DIR, enhancing comprehensive disentanglement. Additionally, our proposed Cyclic Disentanglement Module (CDM) facilitates DSR separation. To refine the process further, we employ the Categorical Features Disentanglement Module (CFDM) to isolate DIR and DSR, coupled with category alignment across scales for improved source-target domain alignment. Given its practical suitability, our model is constructed upon the foundational framework of the Single Shot MultiBox Detector (SSD), which is a one-stage object detection approach. Experimental validation highlights the effectiveness of our method, demonstrating its state-of-the-art performance across three benchmark datasets.

show abstract

RFA-Net: Reconstructed Feature Alignment Network for Domain Adaptation Object Detection in Remote Sensing Imagery

Cited by 18 publications

References 69 publications

A Classwise Vulnerable Part Detection Method for Military Targets

A Classwise Vulnerable Part Detection Method for Military Targets

Domain Adaptation with Contrastive Learning for Object Detection in Satellite Imagery

Triple Feature Disentanglement for One-Stage Adaptive Object Detection

Contact Info

Product

Resources

About