ChangeT$^3$: Evaluating Vision Test-Time Training for Remote Sensing Change Detection

Sharukshan Niranjan; Oshadha Samarakoon; Nichula Wasalathilaka; Roshan Godaliyadda; Parakrama Ekanayake; Vijitha Herath

cs.CVOct 11, 2026

ChangeT $^3$ : Evaluating Vision Test-Time Training for Remote Sensing Change Detection

Authors: Sharukshan Niranjan, Oshadha Samarakoon, Nichula Wasalathilaka, Roshan Godaliyadda, Parakrama Ekanayake, Vijitha Herath

Organizations: University of Peradeniya, Peradeniya, Sri Lanka

Abstract

Bitemporal remote sensing change detection requires representations that preserve spatial detail while capturing contextual information relevant to changes between acquisitions. Visual test-time training offers an alternative to explicit token-to-token attention by adapting a compact inner model to each image. This letter describes ChangeT $^3$ , a vanilla Siamese configuration combining the existing hierarchical ViT $^3$ -Tiny encoder with the multiscale ChangeFormer decoder. The two image streams share learned encoder parameters, while their fast weights are computed independently from their respective spatial features. Each encoder block retains the inherited global simplified SwiGLU and local adaptive depthwise-convolution branches. Temporal comparison is performed by the decoder through learned fusion of paired features at four resolutions, followed by coarse-to-fine aggregation and binary prediction. This separation provides a direct setting for examining image-conditioned inner updates without introducing an additional temporal interaction module. Trained from scratch on LEVIR-CD+ with the same decoder, recipe, and budget as a ChangeFormer control, ChangeT $^3$ reaches an F1 of 80.61 $\pm$ 0.90 against 78.84 $\pm$ 0.21 for ChangeFormer (three seeds each); as in the standard protocol for this dataset, checkpoints are selected on the test split. Trained with both inner updates disabled, the same encoder reaches 78.63 $\pm$ 0.20. Neither model transfers zero-shot to SYSU-CD, DSIFN-CD, or WHU-CD, and ChangeT $^3$ is the slower of the two at small inputs despite similar parameter and operation counts.

Figures & tables

Panel of Figure 1 — Fig. 1: Overview of ChangeT3. (a) A Siamese hierarchical ViT 3 encoder (H-ViT 3 -Tiny) maps the pre- and post-change images to four-scale features with shared persistent weights $\theta$ . Inside every block, each image fits its own temporary fast weights $W^{\star}$ , so $W^{\star}_{A}\neq W^{\star}_{B}$ in general and no information crosses dates in the encoder. (b) The ChangeFormer decoder projects each scale with an MLP shared across dates, concatenates the pair, and learns a temporal difference (ConvDiff: Conv $3{\times}3$ –ReLU–BN–Conv $3{\times}3$ –ReLU); coarser differences are upsampled and added, all scales are fused at $\tfrac{H}{4}$ , and two transposed-conv residual stages recover full resolution. Dashed heads are used only for deep supervision. (c) A TTT layer first writes : one label-free gradient step on $(K,V)$ , with $G=\nabla_{W}\mathcal{L}_{\mathrm{inner}}(W_{0})$ , turns the persistent initialization $W_{0}$ into fast weights $W^{\star}$ (a simplified-SwiGLU model per head and a $3{\times}3$ depthwise kernel; $\eta=1$ , gradient norm taken per head and per kernel), and then reads them with $Q$ . The change-detection loss trains $W_{0}$ and all other persistent weights end to end through the inner step.

Data
Dataset	LEVIR-CD+, 256 $\times$ 256 patches, stride 256
Patches	10,192 train / 5,568 test (val $=$ test)
Scenes	637 train / 348 test, disjoint
Preprocessing	ImageNet normalization; masks binarized at 127; no augmentation
Optimization
Optimizer	AdamW ( $\beta_{1}{=}0.9$ , $\beta_{2}{=}0.999$ ), one parameter group

TABLE I: Training and evaluation settings shared by all runs.

Method	Init.	Params (M)	$P$	$R$	$F_{1}$	IoU	BF1
ChangeFormer (3 seeds)	scratch	41.03	77.03 $\pm$ 1.17	80.76 $\pm$ 0.83	78.84 $\pm$ 0.21	65.07 $\pm$ 0.29	70.58 $\pm$ 0.21
ChangeT 3 (3 seeds)	scratch	40.28	80.54 $\pm$ 1.86	80.70 $\pm$ 0.68	80.61 $\pm$ 0.90	67.52 $\pm$ 1.26	72.69 $\pm$ 0.67
ChangeFormer-P	ADE20K	41.03	84.01	83.95	83.98	72.38	79.30
ChangeT 3 -P	ImageNet	40.28	86.65	83.90	85.25	74.29	78.24
ChangeT 3 -P + layer decay	ImageNet	40.28	87.34	83.38	85.31	74.39	78.47

TABLE II: LEVIR-CD+ results (%). ChangeFormer and ChangeT 3 share decoder, recipe, and budget. Scores are on the test split, which also served for checkpoint selection (test-as-val protocol). Seed rows give mean

\pm

standard deviation. Pretrained rows differ in pretraining data as well as encoder.

Configuration	$\eta_{g}$	$\eta_{\ell}$	$F_{1}$	IoU	Best iter.
Both (vanilla)	1	1	81.64	68.97	43k
Global only	1	0	80.18	66.91	48k
Local only	0	1	80.80	67.79	28k
Neither	0	0	78.73	64.92	38k
Both (3 seeds)	1	1	80.61 $\pm$ 0.90	67.52 $\pm$ 1.26	–
Neither (3 seeds)	0	0	78.63 $\pm$ 0.20	64.79 $\pm$ 0.27	–

TABLE III: Training-time inner-update ablations on LEVIR-CD+ (%). The upper block is seed 42; the lower block gives mean

\pm

standard deviation over seeds. A zero step size keeps the branch and its learned initialization but applies no update. Best iter. is the selected checkpoint.

Fig. 2: Qualitative comparison on the LEVIR-CD+ test split. Samples (a)–(d) are the four highest-ranked test patches from distinct scenes under the selection rule of Section III . For each sample, the upper row shows the change prediction as an error map, where white, black, red and blue denote true positive, true negative, false positive and false negative pixels. The lower row shows structural agreement between the reference contour (green) and the predicted contour (red), with yellow indicating overlap. ChangeFormer and ChangeT 3 share the same decoder and training recipe; they differ only in the encoder and in whether its per-image test-time update is active. The -P variants initialize the encoder from pretrained weights (ADE20K for ChangeFormer, ImageNet for ChangeT 3 ).

	LEVIR-CD+ (in-domain)		SYSU-CD		DSIFN-CD		WHU-CD		Cross-domain
Model	$F_{1}$	IoU	$F_{1}$	IoU	$F_{1}$	IoU	$F_{1}$	IoU	mean $F_{1}$	drop
ChangeFormer (3 seeds)	78.84 $\pm$ 0.21	65.07 $\pm$ 0.29	2.23 $\pm$ 0.32	1.13 $\pm$ 0.17	15.86 $\pm$ 1.33	8.62 $\pm$ 0.79	1.84 $\pm$ 1.08	0.93 $\pm$ 0.55	6.64 $\pm$ 0.40	72.20 $\pm$ 0.62
ChangeT 3 w/o TTT (3 seeds)	78.63 $\pm$ 0.20	64.79 $\pm$ 0.27	0.80 $\pm$ 0.21	0.40 $\pm$ 0.10	7.97 $\pm$ 0.62	4.15 $\pm$ 0.34	1.92 $\pm$ 0.88	0.97 $\pm$ 0.45	3.57 $\pm$ 0.22	75.07 $\pm$ 0.38
ChangeT 3 (3 seeds)	80.61 $\pm$ 0.90	67.52 $\pm$ 1.26	0.87 $\pm$ 0.36	0.44 $\pm$ 0.18	3.41 $\pm$ 0.73	1.73 $\pm$ 0.38	1.22 $\pm$ 0.21	0.61 $\pm$ 0.11	1.83 $\pm$ 0.08	78.78 $\pm$ 0.96
ChangeFormer-P	83.98	72.38	3.52	1.79	18.13	9.97	6.04	3.12	9.23	74.75
ChangeT 3 -P	85.25	74.29	3.86	1.97	11.83	6.29	23.53	13.33	13.07	72.18

TABLE IV: Zero-shot cross-domain evaluation (%) of models trained only on LEVIR-CD+. Seed rows give mean

\pm

standard deviation. Cross-domain mean

F_{1}

averages the three targets; drop is in-domain

F_{1}

minus that mean.

Model	Params (M)	GFLOPs	Latency (ms)	Throughput	Peak mem (GB)	Train	Train mem
			$b{=}1$	pairs/s, $b{=}8$	$b{=}1$ / $b{=}8$	it/s	(GB)
ChangeFormer	41.03	277.5	16.2 $\pm$ 1.3	79.1	0.45 / 2.96	2.81	7.11
ChangeT 3	40.28	284.4	89.9 $\pm$ 3.1	50.4	0.60 / 3.12	1.53	9.55
ChangeT 3 , $\eta{=}0$	40.28	284.4	92.2 $\pm$ 9.0	49.7	0.75 / 3.27	2.33	9.55
ChangeT 3 , update skipped †	–	–	60.2 $\pm$ 3.1	61.9	0.60 / 3.12	–	–

TABLE V: Cost of one forward pass on an image pair at 256

\times

256, fp32, on one RTX 6000 Ada Generation GPU. Latency is mean

\pm

standard deviation over 100 timed passes. † Inner gradients not computed; the output is identical to

\eta{=}0

. The training columns come from the training logs on shared GPUs; contention differed between runs, so they are indicative only and not comparable across rows.

Input	ms	GB	GFLOPs	ms	GB	GFLOPs
	ChangeFormer			ChangeT 3
$256^{2}$	17.3	0.45	277	88.3	0.60	284
$384^{2}$	29.2	0.81	627	103.7	0.96	640
$512^{2}$	48.1	1.31	1120	123.0	1.46	1138
$768^{2}$	110.4	2.74	2556	197.2	2.89	2560
$1024^{2}$	208.6	4.75	4637	345.7	4.90	4551

TABLE VI: Batch-1 latency, peak allocated memory, and operations against input size.

Explore similar work

Feb 21, 2026

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

State Space Models (SSMs) have recently gained traction in remote sensing change detection (CD) for their favorable scaling properties. In this paper, we explore the potential of modern convolutional and attention-based architectures as a competitive alternative. We propose NeXt2Former-CD, an end-to-end framework that integrates a Siamese ConvNeXt encoder initialized with DINOv3 weights, a deformable attention-based temporal fusion module, and a Mask2Former decoder. This design is intended to better tolerate residual co-registration noise and small object-level spatial shifts, as well as semantic ambiguity in bi-temporal imagery. Experiments on LEVIR-CD, WHU-CD, and CDD datasets show that our method achieves the best results among the evaluated methods, improving over recent Mamba-based baselines in both F1 score and IoU. Furthermore, despite a larger parameter count, our model maintains inference latency comparable to SSM-based approaches, suggesting it is practical for high-resolution change detection tasks.

Yufan Wang, Sokratis Makrogiannis, Chandra Kambhamettu

University of South Florida · Delaware State University

Remote Sensing Change Detection Hybrid CNN-Transformer Architectures Convolutional Neural Networks

Ask PDF

Mar 20, 2026

Beyond Quadratic: Linear-Time Change Detection with RWKV

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.

Zhenyu Yang, Gensheng Pei, Tao Chen +4

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

Ask PDF

Jun 20, 2026

CoSA: Correlation-Guided Change Attention with Learnable Residual Gating for Remote Sensing Change Detection

Remote sensing change detection (CD) from bi-temporal imagery is critical for applications such as urban monitoring, disaster assessment, and environmental management, yet robust localization remains challenging under sparse changes, noisy labels, and appearance variations. In this paper, we propose Context Sampling Attention (CoSA), a lightweight decoder-side refinement module that explicitly leverages bi-temporal feature correlation as a control signal for adaptive change-aware feature enhancement. This differs from conventional attention mechanisms that rely on implicit feature weighting without explicit temporal control. In the implemented FC-Siam setting, CoSA computes normalized same-location cross-correlation between paired decoder features, converts low correlation into a change gate, and injects the resulting gated residual at native 1/8 and 1/16 feature scales through learnable residual scaling. This design enables effective discrimination between stable and ambiguous regions without relying on computationally expensive global attention. Extensive experiments on four benchmark datasets (LEVIR-CD, S2Looking, DSIFN, and CLCD) demonstrate consistent improvements over strong baselines, achieving 1.5-2.6% gains in changed-class F1 while introducing negligible parameter overhead. Ablation studies confirm that multiscale placement and learnable residual gating are both important for peak performance. These results indicate that CoSA establishes a practical and effective refinement paradigm for enhancing temporal discriminability in Siamese change detection frameworks.

Abdirashid Omar, Jonghyuk Park

Department of Data Science, Graduate School of Kookmin University Seoul 02707, Republic of Korea

Remote Sensing Change Detection Attention Mechanisms

Ask PDF