CARNet: End-to-End Coupled Association-Recognition Network for Fine-Grained Multi-Drone Collaborative Detection

Zhao Zhang; Ruigang Fu; Yu Zhang; Jiajun Xiang; Zixuan Xiao; Ping Zhong

cs.CVOct 19, 2026

CARNet: End-to-End Coupled Association-Recognition Network for Fine-Grained Multi-Drone Collaborative Detection

Authors: Zhao Zhang, Ruigang Fu, Yu Zhang, Jiajun Xiang, Zixuan Xiao, Ping Zhong

Organizations: College of Electronic Science and Technology, National University of Defense Technology, Changsha, China

Abstract

Multi-drone collaborative detection aims to enhance target recognition by leveraging complementary observations from overlapping aerial viewpoints. Existing approaches typically follow a two-stage pipeline that first establishes cross-view correspondences and then performs fusion-based recognition. However, this decomposition suffers from inconsistent optimization objectives, redundant pairwise modeling, and irreversible error propagation, where incorrect associations inevitably undermine subsequent recognition performance. To overcome these limitations, we propose the Coupled Association-Recognition Network (CARNet), a unified framework that jointly optimizes cross-view target association and recognition fusion through a shared target-pair representation. To improve association reliability in densely populated scenes with visually similar targets, CARNet explicitly incorporates spatial cues into pairwise relation modeling. Moreover, we introduce two complementary causal-inspired intervention strategies: the first enhances cross-view association by emphasizing appearance-confusing pairs, while the second bolsters the robustness of fusion-based recognition by learning stable multi-view representations under incomplete visual evidence. In addition, we present MDMT-CD, a fine-grained extension of the MDMT dataset that enriches the original vehicle annotations with detailed category labels, providing the first benchmark specifically designed for multi-drone collaborative vehicle detection. Extensive experiments demonstrate that CARNet consistently outperforms existing state-of-the-art methods. The MDMT-CD dataset and source code will be available at https://github.com/bigdreaming11/CARNet.

Figures & tables

Figure 1: Overall pipeline of the proposed CARNet. Given two drone views, target features are extracted and used to construct position-aware all-pairs relations. Each target-pair relation is mapped to a unified $C+1$ distribution for joint cross-view association and collaborative recognition, while two causal-inspired intervention are applied during training.

Method	$mAP_{50}$	$mAP_{50-90}$	Latency (ms)
Single Drone	23.45	16.12	–
UVF (AAAI 2026)	16.40	11.00	173
MVCNN (CVPR 2015)	17.39	11.55	88
CMLP	17.96	12.09	103
SMDC (AAAI 2023)	20.99	14.19	94
ARA (AAAI 2025)	21.31	14.41	97

Table 1: Comparison of collaborative detection methods on the MDMT-CD dataset. The best scores are marked in bold.

Figure 2: Visualization of collaborative detection results on the MDMT-CD dataset. Green numbers denote single-drone predictions, and red numbers denote fused predictions. Boxes without red numbers are not associated.

Method	Precision (%)	Recall (%)	MDA (%)	Latency (ms)
ReIDMamba (TMM 2026)	30.07	28.96	14.22	107
Re-ID (ACM MM 2023)	31.53	28.27	15.31	18
OSNet (ICCV 2019)	33.11	30.13	18.11	21
TTS (Drones 2022)	26.97	24.57	12.06	1284
GTA-Net (Drones 2025)	34.64	32.31	19.27	1034
MambaGlue (CVPR 2025)	8.27	8.64	2.87	44

Table 2: Comparison of cross-view object association methods on the MDMT-CD dataset. The best scores are marked in bold.

Figure 3: Visualization of cross-view target association results on the MDMT-CD dataset. Red lines indicate incorrect associations, while yellow lines denote correct associations.

	Method	Precision (%)	Recall (%)	MDA (%)
Visible Light	TTS (Drones 2022)	9.89	18.21	3.48
	GTA-Net (Drones 2025)	16.78	13.23	5.46
	MambaGlue (CVPR 2025)	17.31	14.21	6.33
	MIA-Net (TMM 2023)	29.77	33.21	15.67
	ReIDMamba (TMM 2026)	39.99	43.23	19.86
	OSNet (ICCV 2019)	40.21	46.32	20.76

Table 3: Comparison of Different Methods on M3Track dataset. The best scores are marked in bold.

Figure 4: Visualization of cross-view target association results on the visible-light subset of the M3Track dataset.

Figure 5: Visualization of cross-view target association results on the infrared modality of the M3Track dataset.

Figure 6: Visualization of cross-modal target association results between visible and infrared modalities on the M3Track dataset.

Dataset	Method	MOTA (%)	IDF1 (%)
Wildtrack	KSP-DO(CVPR 2018)	69.6	73.2
	KSP-DO-track(CVPR 2018)	72.2	78.4
	GLMB-DO (TPAMI 2018)	70.1	72.5
	DMCT (Arxiv 2020)	72.8	77.8
	DMCT Stack (Arxiv 2020)	74.6	81.9
	ReST (CVPR 2023)	84.9	86.7

Table 4: Comparison on two natural scenario datasets: Wildtrack and MvMHAT. The best scores are marked in bold.

Method	Position	Strategy 1	Strategy 2	MDA (%)
-			$\checkmark$	23.19
-		$\checkmark$	$\checkmark$	31.80
-	$\checkmark$		$\checkmark$	33.39
CARNet	$\checkmark$	$\checkmark$	$\checkmark$	40.81

Table 5: Ablation study of the proposed modules on the MDMT-CD dataset. “Position” denotes the Position information, “Strategy 1” denotes the Hard-Distinguishing Intervention, and “Strategy 2” denotes the Visual-Evidence Intervention.

Figure 7: Effect of different Drop rates on mAP50.

Method	Backbone	Parameters (M)	GFLOPs (G)	MDA (%)
CARNet	ResNet101	54.88	1294.33	41.04
	ResNet50	24.56	680.18	40.81
	ResNet18	11.83	304.55	35.93
	MobileNetV3	1.35	28.97	29.27
	ShuffleNetV2	0.86	18.41	28.22
LGSGMM	ResNet50	35.88	680.20	39.29

Table 6: Ablation study of different backbones on the MDMT-CD dataset. The input image resolution is

1920\times 1080

Single Detector	mAP 50 (S) (%)	mAP 50 (M) (%)	Gain (%)
YOLOv3-Tiny	18.36	24.15	+5.79
YOLOv3	19.73	24.22	+4.49
YOLOv5-s	22.78	25.03	+2.25
YOLOv8-s	23.45	25.23	+1.78
YOLOv12-s	23.64	25.37	+1.73
YOLOv26-s	23.59	25.33	+1.74

Table 7: Robustness of CARNet to different single-drone detectors.

Aerial Object Detection

Aerial object detection locates and classifies objects in drone and satellite imagery, often under scale, orientation, clutter, and compute constraints.

See all 122 papers

Mar 6, 2026

CollabOD: Collaborative Multi-Backbone with Cross-scale Vision for UAV Small Object Detection

Small object detection in unmanned aerial vehicle (UAV) imagery is challenging because high-altitude viewpoints produce severe scale variation, weak structural cues, and tight computational budgets. Existing lightweight detectors usually fuse multi-scale features after downsampling, where boundary and texture details have already been attenuated and heterogeneous feature streams may be spatially misaligned. To address these issues, we propose CollabOD, a collaborative detection framework that preserves structural details, aligns cross-path features before fusion, and keeps the detection head lightweight at inference time. CollabOD combines a Dual-Path Fusion Stem, a Dense Aggregation Block, a Bilateral Reweighting Module, and a Unified Detail-Aware Head to strengthen localization-oriented representation while limiting extra computation. On VisDrone, CollabOD obtains 52.4 AP50, 30.8 AP75, and 29.9 AP50:95 with 65.5 GFLOPs; on UAVDT it reaches 31.2 AP50 and 17.4 AP50:95; and on AI-TOD it reaches 45.4 AP50 and 20.0 AP50:95 at 137 FPS. The code is available at: https://github.com/Bai-Xuecheng/CollabOD.

Xuecheng Bai, Yuxiang Wang, Chuanzhi Xu +3

Aviation Traffic Control Technology (SHENZHEN) Co., Ltd., Shenzhen, China · The University of Sydney, NSW, Australia · Research Institute of Traffic Control Technology Co., Ltd., Beijing, China +1

Efficient Object Detection Small Object Detection Aerial Object Detection Remote Sensing Small Object Detection Multi-Scale Feature Fusion Real-Time Object Detection+1

Ask PDF

Aug 14, 2026

CoDS: Robust Collaborative Perception via Expert-driven Detection and BEV Segmentation

Collaborative perception breaks through single-view limitations via multi-agent information exchange. However, multi-source noise such as pose errors and communication delays degrades fusion feature quality, constraining perception performance. Joint training of detection and BEV segmentation provides a natural remedy, where segmented road regions help constrain target distributions and detection bounding boxes help recover ambiguous segmentation boundaries. To this end, we propose a robust Collaborative perception framework with expert-driven Detection and bev Segmentation (CoDS). To address spatial inconsistency in fusion quality, we first introduce the Collaborative Reliability Map (CoRM) to explicitly quantify feature quality distribution. Based on CoRM, we design the Semantic Mixture-of-Experts (S-MoE) module to extract differentiated features for inconsistent feature demands. Finally, to further mitigate feature noise degradation, the Bidirectional Task Complementary Interaction (BTCI) refines task-aware features through bidirectional injection. Extensive experiments on OPV2V and V2V4Real datasets show that our CoDS surpasses existing baselines on both tasks and maintains stable robustness under multi-source noise. Code: https://github.com/JinlongW128/CoDS and https://openi.pcl.ac.cn/OpenAIDriving/CoDS.

Jinlong Wang, Yuang Jia, Junhong Lin +2

Guangdong Provincial Key Laboratory of Ultra High Definition Immersive Media Technology, Shenzhen Graduate School Peking University Shenzhen, China · School of Computer Science and Engineering Macau University of Science and Technology Macau, China

Bird's-Eye-View Segmentation Cooperative Perception Autonomous Driving Perception Object Detection Semantic Segmentation

Ask PDF

Mar 13, 2026

GT-Space: Enhancing Heterogeneous Collaborative Perception with Ground Truth Feature Space

In autonomous driving, multi-agent collaborative perception enhances sensing capabilities by enabling agents to share perceptual data. A key challenge lies in handling {\em heterogeneous} features from agents equipped with different sensing modalities or model architectures, which complicates data fusion. Existing approaches often require retraining encoders or designing interpreter modules for pairwise feature alignment, but these solutions are not scalable in practice. To address this, we propose {\em GT-Space}, a flexible and scalable collaborative perception framework for heterogeneous agents. GT-Space constructs a common feature space from ground-truth labels, providing a unified reference for feature alignment. With this shared space, agents only need a single adapter module to project their features, eliminating the need for pairwise interactions with other agents. Furthermore, we design a fusion network trained with contrastive losses across diverse modality combinations. Extensive experiments on simulation datasets (OPV2V and V2XSet) and a real-world dataset (RCooper) demonstrate that GT-Space consistently outperforms baselines in detection accuracy while delivering robust performance. Our code will be released at https://github.com/KingScar/GT-Space.

Wentao Wang, Haoran Xu, Guang Tan

Sun Yat-sen University, Shenzhen Campus, Shenzhen, China · Peng Cheng Laboratory (PCL), Shenzhen, China

Heterogeneous Collaborative Perception Autonomous Driving Perception Representation Alignment Multimodal Sensor Fusion

Ask PDF

Abstract

Aerial Object Detection

Explore similar work

CollabOD: Collaborative Multi-Backbone with Cross-scale Vision for UAV Small Object Detection

CoDS: Robust Collaborative Perception via Expert-driven Detection and BEV Segmentation

GT-Space: Enhancing Heterogeneous Collaborative Perception with Ground Truth Feature Space