Fourier Angle Alignment for Oriented Object Detection in Remote Sensing

Changyu Gu; Linwei Chen; Lin Gu; Ying Fu

cs.CVFeb 27, 2026

Fourier Angle Alignment for Oriented Object Detection in Remote Sensing

Authors: Changyu Gu, Linwei Chen, Lin Gu, Ying Fu

Organizations: Beijing Institute of Technology · University of Hong Kong · The Hong Kong University of Science and Technology · Hong Kong Generative AI Research and Development Center · Tohoku University

Abstract

In remote sensing rotated object detection, mainstream methods suffer from two bottlenecks, directional incoherence at detector neck and task conflict at detecting head. Ulitising fourier rotation equivariance, we introduce Fourier Angle Alignment, which analyses angle information through frequency spectrum and aligns the main direction to a certain orientation. Then we propose two plug and play modules : FAAFusion and FAA Head. FAAFusion works at the detector neck, aligning the main direction of higher-level features to the lower-level features and then fusing them. FAA Head serves as a new detection head, which pre-aligns RoI features to a canonical angle and adds them to the original features before classification and regression. Experiments on DOTA-v1.0, DOTA-v1.5 and HRSC2016 show that our method can greatly improve previous work. Particularly, our method achieves new state-of-the-art results of 78.72% mAP on DOTA-v1.0 and 72.28% mAP on DOTA-v1.5 datasets with single scale training and testing, validating the efficacy of our approach in remote sensing object detection. The code is made publicly available at https://github.com/gcy0423/Fourier-Angle-Alignment .

Figures & tables

Figure 1: The position of FAAFusion and FAA Head.

Panel of Figure 2 — Figure 2: Directional incoherence. High-level feature (c) has strong semantics and large receptive fields but captures the coarse orientation. For example, from (c) we can only infer that the object lies horizontally. Yet many down-samplings blur spatial details, so the orientation signal is low-frequency and vague. Low-level feature (b) keeps rich edges, corners, and textures, giving clear and high-frequency orientation cues.

Figure 3: Task conflict in the detection head. As the orientation varies, the angle predictions are expected to be different while the class predictions are expected to be the same.

Panel of Figure 4 — Figure 4: The top row shows the target images in the spatial domain. The second row gives their matching frequency spectra. In the bottom row, red lines point out the main energy directions in each spectrum. (a) If the spatial picture is rotated, its spectrum distribution will be rotated as well. (b) The main spectral direction of one rectangle-like object is perpendicular to the direction of its major axis.

Figure 5: Structures of FAA, FAAFusion and FAA Head. For FAA, it can be divided into two steps. First, we estimate the main direction in frequent domain. And then FAA accepts an external angle and aligns the main direction with it. For FAAFusion, we use the main direction of the lower local feature as the external angle and align the higher local feature with it. For FAA Head, it aligns the main direction of RoI features with $0\degree$ to obtain the rotation invariant features.

Method	Venue	Backbone	PL	BD	BR	GTF	SV	LV	SH	TC	BC	ST	SBF	RA	HA	SP	HC	mAP
R. F-RCNN [ 37 ]	TPAMI’16	ResNet50	89.40	81.81	47.28	67.44	73.96	73.12	85.03	90.90	85.15	84.90	56.60	64.77	64.70	70.28	62.22	73.17
R3Det [ 45 ]	AAAI’21	ResNet50	89.5	81.2	50.5	66.1	70.9	78.7	78.2	90.8	85.3	84.2	61.8	63.8	68.2	69.8	67.2	73.7
R3Det-GWD [ 46 ]	ICML’21	ResNet50	88.82	82.94	55.63	72.75	78.52	83.10	87.46	90.21	86.36	85.44	64.70	61.41	73.46	76.94	57.38	76.34
R3Det-KLD [ 47 ]	NIPS’21	ResNet50	88.90	84.17	55.80	69.35	78.72	84.08	87.00	89.75	84.32	85.73	64.74	61.80	76.62	78.49	70.89	77.36
S2ANet [ 18 ]	TGRS’21	ResNet50	89.11	82.84	48.37	71.11	78.11	78.39	87.25	90.83	84.90	85.64	60.36	62.60	65.26	69.13	57.94	74.12
RoI Trans. [ 11 ]	CVPR’19	ResNet50	89.01	77.48	51.64	72.07	74.43	77.55	87.76	90.81	79.71	85.27	58.36	64.11	76.50	71.99	54.06	74.05

Table 1: Comparisons with state-of-the-art methods on DOTA-v1.0 [ 12 ] dataset with single scale training and testing.

Method	Backbone	PL	BD	BR	GTF	SV	LV	SH	TC	BC	ST	SBF	RA	HA	SP	HC	CC	mAP
RetinaNet-O [ 29 ]	ResNet50	71.43	77.64	42.12	64.65	44.53	56.79	73.31	90.84	76.02	59.96	46.95	69.24	59.65	64.52	48.06	0.83	59.16
R. F-RCNN [ 37 ]	ResNet50	71.89	74.47	44.45	59.87	51.28	68.98	79.37	90.78	77.38	67.50	47.75	69.72	61.22	65.28	60.47	1.54	62.00
Mask R-CNN [ 20 ]	ResNet50	76.84	73.51	49.90	57.80	51.31	71.34	79.75	90.46	74.21	66.07	46.21	70.61	63.07	64.46	57.81	9.42	62.67
HTC [ 3 ]	-	77.80	73.67	51.40	63.99	51.54	73.31	80.31	90.48	75.12	67.34	48.51	70.63	64.84	64.48	55.87	5.15	63.40
RoI Trans. [ 11 ]	ResNet50	71.70	82.70	53.00	71.50	51.30	74.60	80.60	90.40	78.00	68.30	53.10	73.40	73.90	65.60	56.90	3.00	65.50
O-RCNN [ 44 ]	ResNet50	79.68	81.72	53.36	72.17	52.26	76.47	81.00	90.86	78.93	68.18	65.05	72.17	67.68	65.23	56.11	7.42	66.77

Table 2: Comparisons with state-of-the-art methods on DOTA-v1.5 [ 12 ] dataset with single scale training and testing.

PASCAL VOC2007 [ 13 ]
Method	#Params $\downarrow$	#FLOPs $\downarrow$	AP50 $\uparrow$	AP75 $\uparrow$	mAP $\uparrow$
O-RCNN [ 44 ]	41.13M	134.46G	89.7	79.5	64.77
O-RCNN+ours	63.27M	140.70G	89.8	80.0	66.94
LSKNet [ 25 ]	30.96M	111.42G	90.2	87.9	68.78
LSKNet+ours	48.34M	114.89G	90.6	89.8	70.74
S-RCNN [ 48 ]	45.12M	157.19G	89.5	78.8	69.18

Table 3: Comparisons with three representative state-of-the-art methods on HRSC2016 [ 33 ] dataset. The values of APs are calculated with PASCAL VOC2007 [ 13 ] /VOC2012 [ 14 ] metrics and the mAPs are reported with COCO [ 27 ] styles. All the models are retrained and tested on a single RTX 3090. The Params and FLOPs are calculated with MMRotate [ 55 ] toolbox.

FAAFusion	FAA Head	#Params $\downarrow$	#FLOPs $\downarrow$	mAP $\uparrow$
✘	✘	30.98M	173.68G	77.49
✘	✔	48.35M	177.15G	78.27
✔	✘	32.18M	175.59G	77.91
✔	✔	49.56M	179.06G	78.49

Table 4: Ablation study. We conducted the experiment with LSKNet-S [ 25 ] as the backbone. The Params and FLOPs are calculated with MMRotate [ 55 ] toolbox.

Method	Detecting Head	#Params $\downarrow$	#FLOPs $\downarrow$	mAP $\uparrow$
O-RCNN [ 44 ]	Original Head	41.14M	211.43G	75.81
	Strip Head	55.82M	258.35G	76.11
	FAA Head	58.51M	214.90G	76.18
LSKNet [ 25 ]	Original Head	30.98M	173.68G	77.49
	Strip Head	45.65M	220.60G	78.04
	FAA Head	48.35M	177.15G	78.27

Table 5: Comparison with other detecting head on other remote sensing object detectors. The Params and FLOPs are calculated with MMRotate [ 55 ] toolbox.

Panel of Figure 6 — Figure 6: Effectiveness of our method at Oriented R-CNN [ 44 ] and LSKNet [ 25 ] compared to the original methods on HRSC2016 [ 33 ] dataset. In the high IoU threshold range, our method performs better. The AP values are calculated with PASCAL VOC2007 [ 13 ] and VOC2012 [ 14 ] metrics.

Explore similar work

RMK RetinaNet: Rotated Multi-Kernel RetinaNet for Robust Oriented Object Detection in Remote Sensing Imagery
Mar 5, 2026Huiran SunObject Detection Aerial Object Detection
School of Computer Science and Engineering, Changchun University of Technology, 2055 Yan’an Street, Chaoyang District, Changchun, 130012, Jilin Province, China
FGAA-FPN: Foreground-Guided Angle-Aware Feature Pyramid Network for Oriented Object Detection
Feb 11, 2026Jialin MaMulti-Scale Feature Fusion Object Detection
International school of BUPT, Beijing University Of Posts And Telecommunications, Beijing,China.
Real-Time Oriented Object Detection Transformer in Remote Sensing Images
Mar 16, 2026Zeyu Ding, Yong Zhou, Jiaqi Zhao +4Object Detection Detection Transformer
School of Computer Science and Technology / School of Artificial Intelligence, China University of Mining and Technology, Xuzhou 221116, China · Mine Digitization Engineering Research Center of the Ministry of Education, Xuzhou 221116, China · Jiangsu Provincial Industrial Technology Engineering Center for Intelligent Sensing and Emergency IoT in Underground Space, Xuzhou 221116, China +1

Mar 5, 2026J/K move · Enter open · S save

RMK RetinaNet: Rotated Multi-Kernel RetinaNet for Robust Oriented Object Detection in Remote Sensing Imagery

Huiran Sun

School of Computer Science and Engineering, Changchun University of Technology, 2055 Yan’an Street, Chaoyang District, Changchun, 130012, Jilin Province, China

Object Detection Aerial Object Detection Multi-Scale Feature Fusion

Open paper AskPDF arXiv

Rotated object detection in remote sensing imagery is hindered by three major bottlenecks: non-adaptive receptive field utilization, inadequate long-range multi-scale feature fusion, and discontinuities in angle regression. To address these issues, we propose Rotated Multi-Kernel RetinaNet (RMK RetinaNet). First, we design a Multi-Scale Kernel (MSK) Block to strengthen adaptive multi-scale feature extraction. Second, we incorporate a Multi-Directional Contextual Anchor Attention (MDCAA) mechanism into the feature pyramid to enhance contextual modeling across scales and orientations. Third, we introduce a Bottom-up Path to preserve fine-grained spatial details that are often degraded during downsampling. Finally, we develop an Euler Angle Encoding Module (EAEM) to enable continuous and stable angle regression. Extensive experiments on DOTA-v1.0, HRSC2016, and UCAS-AOD show that RMK RetinaNet achieves performance comparable to state-of-the-art rotated object detectors while improving robustness in multi-scale and multi-orientation scenarios.

Abstract

Explore similar work

RMK RetinaNet: Rotated Multi-Kernel RetinaNet for Robust Oriented Object Detection in Remote Sensing Imagery

FGAA-FPN: Foreground-Guided Angle-Aware Feature Pyramid Network for Oriented Object Detection

Real-Time Oriented Object Detection Transformer in Remote Sensing Images