Data augmentation has shown significant advancements in computer vision to improve model performance over the years, particularly in scenarios with limited and insufficient data. Currently, most studies focus on adjusting the image or its features to expand the size, quality, and variety of samples during training in various tasks including object detection. However, we argue that it is necessary to investigate bounding box transformations as a data augmentation technique rather than image-level transformations, especially in aerial imagery due to potentially inconsistent bounding box annotations. Hence, this letter presents a thorough investigation of bounding box transformation in terms of scaling, rotation, and translation for remote sensing object detection. We call this augmentation strategy NBBOX (Noise Injection into Bounding Box). We conduct extensive experiments on DOTA and DIOR-R, both well-known datasets that include a variety of rotated generic objects in aerial images. Experimental results show that our approach significantly improves remote sensing object detection without whistles and bells and it is more time-efficient than other state-of-the-art augmentation strategies.

Labeling errors in remote sensing (RS) image segmentation datasets often remain implicit and subtle due to ambiguous class boundaries, mixed pixels, shadows, complex terrain features, and subjective annotator bias. Furthermore, the scarcity of annotated RS data due to the high cost of labeling complicates training noise-robust models. While sophisticated mechanisms such as label selection or noise correction might address the issue mentioned above, they tend to increase training time and add implementation complexity. In this paper, we propose NSegment—a simple yet effective data augmentation solution to mitigate this issue. Unlike traditional methods, it applies elastic transformations only to segmentation labels, varying deformation intensity per sample in each training epoch to address annotation inconsistencies. Experimental results demonstrate that our approach improves the performance of RS image segmentation over various state-of-the-art models.

Recently, deep learning models have achieved great success in computer vision applications, relying on large-scale class-balanced datasets. However, imbalanced class distributions still limit the wide applicability of these models due to degradation in performance. To solve this problem, in this paper, we concentrate on the study of cross entropy which mostly ignores output scores on incorrect classes. This work discovers that neutralizing predicted probabilities on incorrect classes improves the prediction accuracy for imbalanced image classification. This paper proposes a simple but effective loss named complement cross entropy based on this finding. The proposed loss makes the ground truth class overwhelm the other classes in terms of softmax probability, by neutralizing probabilities of incorrect classes, without additional training procedures. Along with it, this loss facilitates the models to learn key information especially from samples on minority classes. It ensures more accurate and robust classification results on imbalanced distributions. Extensive experiments on imbalanced datasets demonstrate the effectiveness of the proposed method.

최근 확산 모델은 텍스트 프롬프트로부터 이미지를 생성하는 핵심 기술로 상당한 주목을 받고 있다. 이러한 모델은 특히 높은 정확도로 고해상도 이미지를 생성하는 능력으로 인정받고 있다. 또 하나 주목할 만한 특징은 확산 모델의 유연성으로, 추가 학습을 통해 특정 요구 사항에 맞게 쉽게 미세 조정할 수 있다는 점이다. 이러한 적응력 덕분에 확산 모델은 다양한 분야에 적용되고 있으며, 그 중 하나가 데이터 증강이다. 본 연구에서는 객체 탐지 모델의 학습 효율성을 높이는 새로운 데이터 증강 기법을 제안한다. 구체적으로, 본 접근법은 바운딩 박스를 입력으로 받아 SAR(합성 개구 레이더) 영상을 생성하는 확산 모델 기반의 조건부 생성 기법을 활용한다. 이 혁신적인 방법은 확산 모델이 객체 탐지 성능을 효과적으로 향상시킬 수 있음을 보여준다.

In the rapidly evolving field of remote sensing (RS), the need for efficient and accurate scene classification is paramount. RS imagery comprising satellite and aerial imagery often faces challenges such as varying scales and diverse environmental conditions, which can significantly affect the discernibility of important features. To address these challenges, this article introduces a lightweight dual-branch network architecture that adequately handles scale variations and complex scene compositions. The first branch, progressive feature processing branch (PFPB), of the proposed framework is engineered to extract rich multiple-scale features through collaborative parallel stages and intrabranch and interbranch connectivity with optimized computational resources. The second branch, InXformer branch (IXB) enhances the system’s capability to assimilate global context and long-range dependencies essential for comprehensive scene analysis utilizing an involution-based transformer approach. Experimental validation in three challenging datasets sourced from diverse aerial platforms demonstrates the greater effectiveness of the proposed network. The proposed network achieves a weighted ${F}1$ of 97.15% in the AIDERSv2 dataset, surpassing other methods such as DecoupleNet by more than 2%, while maintaining high efficiency with 0.41M parameters, lower computational overhead with 0.96 GFLOPs, and a higher processing speed of 4616 frames/s (FPS). With regards to WHU-RS19 and UCM datasets, the devised network achieves 93.69% and 94.57% weighted- ${F}1$ score, respectively. These results underscore the ability of the proposed network to efficiently handle diverse scene compositions by delivering state-of-the-art performance.