A comprehensive and trustworthy benchmark of AI methods for change detection in Earth observation

Tadej Tomanič; Alice Baudhuin; Jan Sotošek; Jure Brence; Panče Panov; Nikola Simidjievski; Dragi Kocev

cs.CVAug 28, 2026

A comprehensive and trustworthy benchmark of AI methods for change detection in Earth observation

Authors: Tadej Tomanič, Alice Baudhuin, Jan Sotošek, Jure Brence, Panče Panov, Nikola Simidjievski, Dragi Kocev

Abstract

Change detection in Earth observation (EO) is critical for monitoring land surface transformations, yet recent research in the field is constrained by inconsistent evaluation protocols and a narrow focus on predictive accuracy without regard for computational efficiency. To address this, we present a standardized, open-source benchmark for evaluating state-of-the-art (SOTA) deep learning methods for Earth observation change detection. We conduct a comprehensive analysis of ten representative model architectures, ranging from convolutional networks (CNNs) to vision transformers (ViTs), across ten heterogeneous change detection datasets. We rigorously evaluate these models with identical experimental protocols, comparing models trained from scratch against those utilizing pre-trained weights. Furthermore, we evaluate predictive performance alongside computational efficiency, including parameter counts and inference latency. Our findings reveal that well-optimized classical architectures, such as Siamese U-Nets, frequently outperform more complex contemporary models when computational efficiency is factored in, and that pre-training consistently provides a significant performance boost with no additional inference cost. To ensure complete transparency and reproducibility, all experimental resources, including standardized data splits, training scripts, training logs, and model checkpoints are publicly available and adhere to FAIR principles (Findable, Accessible, Interoperable, and Reusable).

Figures & tables

Dataset	Image source	Image pairs	Bands	Patch size	GSD/ Resolution
BANDON ( Pang et al., 2023 )	Off-nadir aerial	2,283	3	$2048\times 2048$	$0.6$ m
CLCD ( Liu et al., 2022 )	GaoFen-2	600	3	$512\times 512$	$0.5$ – $2$ m
DSIFN ( Zhang et al., 2020 )	Google Earth	3,988	3	$512\times 512$	$0.5$ – $2$ m
EGY-BCD ( Holail et al., 2023 )	Google Earth	6,091	3	$256\times 256$	$0.25$ m
LEVIR-CD+ ( Chen and Shi, 2020 )	Google Earth	985	3	$1024\times 1024$	$0.5$ m
MSBC ( Li et al., 2022 )	GaoFen-2	3,769	3	$256\times 256$	$0.8$ – $10$ m

Table 1: Key characteristics of the change detection datasets used for the benchmarking experiments.

Figure 1: Example image pair ( $T_{1},T_{2}$ ) and ground truth change mask from the Season-Varying CDD dataset ( Lebedev et al., 2018 ) .

Model	Year	Params (M)	GFLOPs
BIT ( Chen et al., 2022 )	2021	3.50	10.88
CGNet ( Han et al., 2023 )	2023	33.68	87.55
ChangeFormerV6 ( Bandara and Patel, 2022 )	2022	41.03	138.77
ChangeViT ( Zhu et al., 2026 )	2026	20.66	26.36
CSSM ( Ghazaei and Aptoula, 2026 )	2025	2.59	2.78
HRNet SiamConc ( Sun et al., 2019 )	2019	21.48	9.55

Table 2: Summary of the representative model architectures considered in this benchmark. FLOPs and parameter counts are reported for bi-temporal input tensor pairs of size

2\times 3\times 256\times 256

Figure 2: Ontological schema used by the benchmark.

(a) Trained from scratch
Dataset	BIT	CGNet	ChangeFormerV6	ChangeViT	CSSM	HRNet SiamConc	SiamCRNN	STANet	TinyCD	U-Net SiamConc	Average
BANDON	0.5015	0.4785	0.5661	0.5545	0.4787	0.5231	0.5511	0.4921	0.5306	0.4840	0.5160
CLCD	0.6920	0.6593	0.6569	0.6637	0.4596	0.6375	0.6950	0.6120	0.6612	0.7048	0.6442
DSIFN	0.7711	0.8109	0.7453	0.7919	0.6829	0.6409	0.7863	0.7224	0.7133	0.7593	0.7424
EGY-BCD	0.7783	0.8214	0.7594	0.7770	0.7005	0.7619	0.7929	0.7598	0.7619	0.7762	0.7689
LEVIR-CD+	0.6378	0.6412	0.6561	0.7181	0.5785	0.6766	0.6907	0.5449	0.7201	0.7201	0.6584

Table 3: Mean Intersection over Union (mIoU) benchmark results across ten change detection datasets, comparing models trained from scratch versus pre-trained backbones. The best performance per dataset within each training setup is indicated in bold , and the second-best is underlined .

Figure 3: Comparison of a) mIoU and b) macro $F_{1}$ -score for the benchmarked change detection models. Points represent the metric values averaged across the ten benchmark datasets for each architecture. Error bars denote the 95% confidence interval, indicating the performance variance across the datasets. The models are evaluated under two different training regimes: trained from scratch (orange circles, dashed line) and initialized with pre-trained weights (blue squares, solid line). Figures c) and d) present diverging bar charts illustrating the absolute performance difference ( $\Delta$ ) between the pre-trained models and models trained from scratch. Blue bars extending to the right indicate a performance gain from pre-training, while orange bars to the left indicate a degradation. Models lacking pre-trained weights (ChangeFormerV6 and CSSM) are excluded from the delta comparisons.

Model	Activation	Latency	Throughput	Average train time (h)
	memory (GB)	(ms)	(FPS)	Scratch	Pre-trained
BIT	8.39	11.92	83.89	2.21 $\pm$ 2.57	1.72 $\pm$ 2.06
CGNet	11.51	6.02	166.11	0.93 $\pm$ 1.02	0.15 $\pm$ 0.18
ChangeFormerV6	14.78	22.32	44.80	1.53 $\pm$ 1.68	/
ChangeViT	10.40	24.47	40.87	2.12 $\pm$ 2.40	1.58 $\pm$ 1.81
CSSM	1.64	60.63	16.49	1.79 $\pm$ 2.13	/

Table 4: Resource profiling and computational efficiency of the benchmarked models. Peak activation memory is estimated for a training batch of size 16. Latency and throughput are measured on a single GPU with bi-temporal input tensor of size

2\times 3\times 256\times 256

(batch size 1 for inference). Training times represent the average wall-clock duration to reach convergence across all benchmark datasets.

Figure 4: Trade-off between predictive performance and computational efficiency for the benchmarked change detection models. a) Theoretical efficiency displaying average mIoU against GFLOPs, with shape size representing the total number of parameters (in millions). b) Practical hardware efficiency, showcasing average mIoU against inference latency (ms), with bubble size indicating peak activation memory (GB). Marker colors correspond to the specific model architectures, while marker shapes denote the training type: random initialization from scratch (circles) and ImageNet-1K pre-training (squares). Models positioned toward the top-left with smaller shape sizes represent the optimal model architecture, yielding high performance at low resource costs.

Figure 5: Qualitative comparison of model predictions on the Season-varying CDD dataset. The figure displays the bi-temporal input imagery, ground truth masks, and predicted change maps for three models trained from scratch: a) ChangeViT, b) U-Net SiamConc, and c) CGNet.

Figure 6: Qualitative comparison of model predictions (flood mapping) on the OMBRIA dataset. The figure displays the bi-temporal input imagery, ground truth masks, and predicted change maps for three models trained from scratch: a) ChangeFormerV6, b) U-Net SiamConc, and c) STANet.

Remote Sensing Change Detection

Remote sensing change detection finds and interprets differences across Earth-observation images acquired at different times.

See all 89 papers

Beyond Quadratic: Linear-Time Change Detection with RWKV
Mar 20, 2026Zhenyu Yang, Gensheng Pei, Tao Chen +4Remote Sensing Change Detection
School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China · State Key Laboratory of Intelligent Manufacturing of Advanced Construction Machinery, China · Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, Korea
Decoding the Delta: Unifying Remote Sensing Change Detection and Understanding with Multimodal Large Language Models
Apr 15, 2026Xiaohe Li, Jiahao Li, Kaixin Zhang +5Remote Sensing Change Detection Remote Sensing VQA
Beijing, China Aerospace Information Research Institute, CAS · Beijing, China Space Engineering University · Beijing, China Capital Normal University
NeXt2Former-CD: Efficient Remote Sensing Change Detection with Modern Vision Architectures
Feb 21, 2026Yufan Wang, Sokratis Makrogiannis, Chandra KambhamettuRemote Sensing Change Detection Hybrid CNN-Transformer Architectures
University of South Florida · Delaware State University

Mar 20, 2026J/K move · Enter open · S save

Beyond Quadratic: Linear-Time Change Detection with RWKV

Zhenyu Yang, Gensheng Pei, Tao Chen, Xia Yuan+3

School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China · State Key Laboratory of Intelligent Manufacturing of Advanced Construction Machinery, China · Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, Korea

Remote Sensing Change Detection

Open paper AskPDF arXiv

Existing paradigms for remote sensing change detection are caught in a trade-off: CNNs excel at efficiency but lack global context, while Transformers capture long-range dependencies at a prohibitive computational cost. This paper introduces ChangeRWKV, a new architecture that reconciles this conflict. By building upon the Receptance Weighted Key Value (RWKV) framework, our ChangeRWKV uniquely combines the parallelizable training of Transformers with the linear-time inference of RNNs. Our approach core features two key innovations: a hierarchical RWKV encoder that builds multi-resolution feature representation, and a novel Spatial-Temporal Fusion Module (STFM) engineered to resolve spatial misalignments across scales while distilling fine-grained temporal discrepancies. ChangeRWKV not only achieves state-of-the-art performance on the LEVIR-CD benchmark, with an 85.46% IoU and 92.16% F1 score, but does so while drastically reducing parameters and FLOPs compared to previous leading methods. This work demonstrates a new, efficient, and powerful paradigm for operational-scale change detection. Our code and model are publicly available.

Abstract

Remote Sensing Change Detection

Explore similar work

Beyond Quadratic: Linear-Time Change Detection with RWKV

Decoding the Delta: Unifying Remote Sensing Change Detection and Understanding with Multimodal Large Language Models

NeXt2Former-CD: Efficient Remote Sensing Change Detection with Modern Vision Architectures