Cross-Domain Collaborative Enhancement for Tiny Object Detection in Remote Sensing Images

ZHANG Tianyang; ZHANG Xiangrong; WANG Guanchun; TANG Xu

doi:10.11999/JEIT260317

Article Contents

Article Navigation > Journal of Electronics & Information Technology > 2026 >

ZHANG Tianyang, ZHANG Xiangrong, WANG Guanchun, TANG Xu. Cross-Domain Collaborative Enhancement for Tiny Object Detection in Remote Sensing Images[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT260317

Citation:

ZHANG Tianyang, ZHANG Xiangrong, WANG Guanchun, TANG Xu. Cross-Domain Collaborative Enhancement for Tiny Object Detection in Remote Sensing Images[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT260317

Citation:

PDF( 4816 KB)

Cross-Domain Collaborative Enhancement for Tiny Object Detection in Remote Sensing Images

doi: 10.11999/JEIT260317 cstr: 32379.14.JEIT260317

School of Artificial Intelligence, Xidian University, Xi’an 710126, China

Funds: Natural Science Foundation of China (62501433, 62506285, 62571387), Fundamental Research Funds for the Central Universities under Grant QTZX25070, Natural Science Basic Research Plan in Shaanxi Province of China (2025JC-YBQN-795)

Received Date: 2026-03-18
Accepted Date: 2026-07-06
Rev Recd Date: 2026-04-24

Available Online: 2026-07-19

Abstract

Abstract

Objective The rapid development of deep learning has significantly advanced object detection in remote sensing images (RSIs).Due to constraints imposed by imaging conditions and object size, a large number of tiny objects (with a pixel area of less than 16×16) are widely distributed in RSIs.However, compared with satisfactory detection performance on normal-scaled objects, current object detection methods exhibit a significant performance gap for tiny objects.This is primarily attributed to the two critical limitations: insufficient positive label assignment and weak feature representations.To address these issues, a novel Cross-Domain Collaborative Enhancement Detector (CDCEDet) is proposed.The CDCEDet jointly optimizes the label assignment in the spatial domain and strengthens the feature representations in the frequency domain for tiny objects, thereby establishing an accurate and robust framework for remote sensing tiny object detection. Methods The overall framework of the proposed CDCEDet is depicted in Fig.2 and comprises three key components.First, a Scale-Adaptive Anchor Generator (SAAG) is devised to dynamically generate anchors matched with the scales of ground-truth (GT) objects, effectively mitigating the scale mismatch issue neglected in previous works.Compared to conventional uniform-scale anchor generators, the proposed SAAG significantly increases positive samples for tiny objects, even under the IoU-threshold-based label assignment.Second, a Quantile-based Adaptive Label Assignment (QALA) mechanism is proposed to replace the fixed IoU threshold label assignment.The QALA exploits the IoU distribution between each GT object and its matched anchors to dynamically generate an adaptive threshold tailored to each GT object, thereby further boosting positive samples for tiny objects.Third, a Frequency-Adaptive Fusion (FAF) module is constructed to enhance feature representations from a frequency-domain perspective.On the one hand, adaptive high-pass filters are employed to highlight high-frequency information, which mitigates the information loss during channel compression.On the other hand, adaptive low-pass filters are applied to smooth and up-sample high-level features, thereby alleviating the feature inconsistency within the up-sampled objects. Results and Discussions Extensive experiments are conducted on two publicly available remote sensing tiny object detection datasets, AI-TODv2 and AI-TOD-R, by comparing the proposed method with several latest approaches, including RFLA, DCNet, and DFCL.On the AI-TODv2 dataset (Table 1), the proposed CDCEDet achieves performance gains of 1.8% and 0.7% in terms of AP₅₀ and AP_50-95, respectively, compared to the latest method.On the AI-TOD-R dataset (Table 2), AP₅₀ and AP_50-95 are improved by 2.6% and 0.7%, respectively.These results demonstrate that the proposed CDCEDet possesses superior performance and robust generalization capabilities for tiny object detection in RSIs.Ablation studies and configuration analyses of the CDCEDet core modules (Table 3, Table 4, Table 5, and Table 6) further verify the effectiveness of each proposed component and their complementary nature.Qualitative evaluations on both datasets (Fig.3) confirm that the proposed method accurately detects tiny objects in both sparse and dense distributions scenarios.Fig.4 confirms that the proposed SAAG generates scale-matched anchors for each object and assigns more positive samples to tiny objects than widely used uniformly distributed anchor generator.Furthermore, in comparative visualizations with RFLA and DCNet (Fig.5), CDCEDet exhibits superior detection accuracy and demonstrates a more effective reduction in missed detections. Conclusions This paper proposes a novel CDCEDet to address the insufficient label assignment and weak feature representations in remote sensing tiny object detection.Specifically, a SAAG is proposed to generate scale-matched anchors tailored to each GT object, significantly increasing the number of positive samples for tiny objects.Additionally, a QALA mechanism is devised that exploits the IoU distribution between GT objects and their matched anchors to dynamically adjust the assignment threshold, thereby effectively mitigating the scale bias induced by the fixed threshold.Finally, a FAF module is constructed from a frequency-domain perspective, which adopts the adaptive high-pass filters and low-pass filters to strengthen the feature representations for tiny objects.Extensive experiments on two tiny object detection benchmarks demonstrate the superior performance and robust generalization of the proposed CDCEDet.Future research will focus on enhancing model efficiency and real-time performance to facilitate the practical deployment of the proposed approach in remote sensing scenarios.
- Tiny object detection,
- Remote sensing images,
- Label assignment,
- Feature fusion,
- Frequency learning

FullText(HTML)

References(24)

References

[1]	邹旻瑞, 李宇轩, 戴一冕, 等. UMM-Det: 面向异构多模态遥感影像的一体化目标检测框架[J]. 电子与信息学报, 2025, 47(12): 4704–4713. doi: 10.11999/JEIT250933. ZOU Minrui, LI Yuxuan, DAI Yimian, et al. UMM-Det: A unified object detection framework for heterogeneous multimodal remote sensing imagery[J]. Journal of Electronics & Information Technology, 2025, 47(12): 4704–4713. doi: 10.11999/JEIT250933.
[2]	ZHANG Xiangrong, ZHANG Tianyang, WANG Guanchun, et al. Remote sensing object detection meets deep learning: A metareview of challenges and advances[J]. IEEE Geoscience and Remote Sensing Magazine, 2023, 11(4): 8–44. doi: 10.1109/MGRS.2023.3312347.
[3]	XU Chang, WANG Jinwang, YANG Wen, et al. Detecting tiny objects in aerial images: A normalized Wasserstein distance and a new benchmark[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2022, 190: 79–93. doi: 10.1016/j.isprsjprs.2022.06.002.
[4]	FU Ronghao, CHEN Chengcheng, YAN Shuang, et al. Gaussian similarity-based adaptive dynamic label assignment for tiny object detection[J]. Neurocomputing, 2023, 543: 126285. doi: 10.1016/J.NEUCOM.2023.126285.
[5]	XU Chang, WANG Jinwang, YANG Wen, et al. RFLA: Gaussian receptive field based label assignment for tiny object detection[C]. 17th European Conference on Computer Vision-ECCV 2022, Tel Aviv, Israel, 2022: 526–543.
[6]	刘杰, 刘书豪, 田明, 等. 复杂环境下无人机航拍小目标检测算法[J]. 电子与信息学报, 2026, 48(4): 1763–1773. LIN T Y, DOLLÁR P, GIRSHICK R B, et al. Feature pyramid networks for object detection[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, USA, 2017: 936–944. LIU Jie, LIU Shuhao, TIAN Ming, et al. Small object detection algorithm for UAV aerial images in complex environments[J]. Journal of Electronics & Information Technology, 2026, 48(4): 1763–1773. doi: 10.11999/JEIT251126. .
[7]	王坤, 丁麒龙. 利用自适应融合和混合锚检测器的遥感图像小目标检测算法[J]. 电子与信息学报, 2024, 46(7): 2942–2951. doi: 10.11999/JEIT230966. WANG Kun and DING Qilong. Remote sensing images small object detection algorithm with adaptive fusion and hybrid anchor detector[J]. Journal of Electronics & Information Technology, 2024, 46(7): 2942–2951. doi: 10.11999/JEIT230966.
[8]	WU Jixiang, PAN Zongxu, LEI Bin, et al. FSANet: Feature-and-spatial-aligned network for tiny object detection in remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5630717. doi: 10.1109/TGRS.2022.3205052.
[9]	CHEN Linwei, FU Ying, GU Lin, et al. Frequency-aware feature fusion for dense image prediction[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024, 46(12): 10763–10780. doi: 10.1109/TPAMI.2024.3449959.
[10]	REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137–1149. doi: 10.1109/TPAMI.2016.2577031.
[11]	WANG Jiaqi, CHEN Kai, YANG Shuo, et al. Region proposal by guided anchoring[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, USA, 2019: 2960–2969. doi: 10.1109/CVPR.2019.00308.
[12]	MAGID S A, ZHANG Yulun, WEI Donglai, et al. Dynamic high-pass filtering and multi-spectral attention for image super-resolution[C]. 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, Canada, 2021: 4288–4297. doi: 10.1109/ICCV48922.2021.00425.
[13]	XU Chang, ZHANG Ruixiang, YANG Wen, et al. Oriented tiny object detection: A dataset, benchmark, and dynamic unbiased learning[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026, 48(3): 3167–3184. doi: 10.1109/TPAMI.2025.3634161.
[14]	HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, USA, 2016: 770–778. doi: 10.1109/CVPR.2016.90.
[15]	LIU Dongyang, ZHANG Junping, QI Yunxiao, et al. Exploring lightweight structures for tiny object detection in remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 5623215. doi: 10.1109/TGRS.2025.3567345.
[16]	ZHANG Tianyang, ZHANG Xiangrong, ZHU Xiaoqian, et al. Multistage enhancement network for tiny object detection in remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5611512. doi: 10.1109/TGRS.2024.3363614.
[17]	HU Xinkai, REN Zhida, BHATTI U A, et al. DCEDet: Tiny object detection in remote sensing images based on dual-contrast feature enhancement and dynamic distance measurement[J]. Remote Sensing, 2025, 17(16): 2876. doi: 10.3390/rs17162876.
[18]	ZHANG Tianyang, ZHANG Xiangrong, HUA Chaozhuo, et al. Tiny object detection based on dynamic scale-awareness label assignment and contextual enhancement[J]. Pattern Recognition, 2026, 172: 112449. doi: 10.1016/j.patcog.2025.112449.
[19]	BIAN Jinghao, FENG Mingtao, DONG Weisheng, et al. Feature information driven position Gaussian distribution estimation for tiny object detection[C]. 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, USA, 2025: 30376–30386. doi: 10.1109/CVPR52734.2025.02828.
[20]	DING Jian, XUE Nan, LONG Yang, et al. Learning RoI transformer for oriented object Detection in aerial images[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, USA, 2019: 2849–2858. doi: 10.1109/CVPR.2019.00296.
[21]	XIE Xingxing, CHENG Gong, WANG Jiabao, et al. Oriented R-CNN for object detection[C]. 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, Canada, 2021: 3520–3529. doi: 10.1109/ICCV48922.2021.00350.
[22]	HAN Jiaming, DING Jian, XUE Nan, et al. ReDet: A rotation-equivariant detector for aerial object detection[C]. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, USA, 2021: 2785–2794. doi: 10.1109/CVPR46437.2021.00281.
[23]	LI Wenting, CHEN Yijie, HU Xuankai, et al. Oriented RepPoints for aerial object detection[C]. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, USA, 2022: 1819–1828. doi: 10.1109/CVPR52688.2022.00187.
[24]	ZENG Ying, CHEN Yushi, YANG Xue, et al. ARS-DETR: Aspect ratio-sensitive detection transformer for aerial oriented object detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5610315. doi: 10.1109/TGRS.2024.3364713.