---
title: 'Seg2Change: OVCD Adapter for Remote Sensing'
url: https://www.emergentmind.com/topics/seg2change
type: topic
---

# Seg2Change: OVCD Adapter for Remote Sensing

Seg2Change is an adapter for open-vocabulary change detection (OVCD) in remote sensing that adapts open-vocabulary semantic segmentation models to bi-temporal change analysis by decoupling dense, category-agnostic change localization from semantic recognition. Its central design is to detect **where** a change occurs with a category-agnostic change head and to determine **what** changed with an open-vocabulary semantic segmentation model. The framework is introduced together with **CA-CDD**, a category-agnostic change detection dataset, and is reported to achieve state-of-the-art OVCD performance, including **+9.52 IoU on WHU-CD** and **+5.50 mIoU on SECOND** [2604.11231].

## 1. OVCD formulation and the problem Seg2Change addresses

Seg2Change is defined in the setting of remote-sensing OVCD, where changes between bi-temporal images are described by arbitrary textual or semantic labels rather than a predefined closed set. In this formulation, conventional change detection pipelines are inadequate because they are typically limited to predefined classes in training datasets, which constrains scalability in real-world scenarios. The framework therefore treats OVCD as a joint vision-language problem in which semantic recognition must remain open-vocabulary while change localization must remain reliable across heterogeneous categories [2604.11231].

The work explicitly positions itself against two instance-based OVCD paradigms. The first is **M–C–I**—**Mask proposal → Comparator → Identifier**—and the second is **I–M–C**—**Identifier → Mask proposal → Comparator**. Both are described as vulnerable to error propagation from mask generation, brittle fixed-threshold decision rules, lower parallelizability because proposals are processed instance by instance, and task-dependent paradigm switching. Seg2Change replaces these proposal-centric pipelines with direct dense prediction of a category-agnostic change map, thereby removing proposal generation and threshold-based instance classification from the primary localization path.

A recurrent misconception is that open-vocabulary change detection can be solved by applying an open-vocabulary segmenter independently at two times and comparing the outputs post hoc. Seg2Change rejects that assumption. Its architecture treats category-agnostic localization as a distinct subproblem and introduces a dedicated change head for it, rather than asking the semantic segmenter to solve localization implicitly.

## 2. CA-CDD and category-agnostic supervision

A major component of Seg2Change is **CA-CDD (Category-Agnostic Change Detection Dataset)**, which is constructed by re-annotating coarse change labels from existing datasets so that change regions are no longer limited to a small set of predefined semantic categories. CA-CDD is derived from the training splits of **SECOND**, **JL1-CD**, and **CNAM-CD**. Instead of preserving only category-restricted change definitions, the re-annotation marks regions where the category changes between time \(t_1\) and time \(t_2\) as change regardless of which specific classes are involved [2604.11231].

This supervision is used to train the category-agnostic change head. The stated rationale is that CA-CDD is closer to real OVCD because it supports arbitrary category transitions rather than only building change, land-cover subsets, or a fixed semantic inventory. The paper also specifies that the **SECOND test set is not included in CA-CDD**, which is important for avoiding leakage in evaluation.

The dataset has methodological significance beyond its immediate training role. It recasts the supervision target from “detect change among known categories” to “detect category transition itself.” This suggests a different inductive bias: the model is encouraged to learn change evidence that is semantically neutral at the localization stage and only semantically specific at the indexing stage.

## 3. Architecture: category-agnostic localization and semantic indexing

Seg2Change has two branches. The first is the **Category-Agnostic Change Head (CACH)**, which predicts a category-agnostic change map \(M_{\text{ca}}\). The second is an **Open-Vocabulary Semantic Segmentation (OVSS)** model, which takes the same bi-temporal images together with text prompts and outputs semantic segmentation maps \(M^1\) and \(M^2\). The two outputs are combined as
\[
M_{\text{ch}}^t = M_{\text{ca}} \cdot M^t
\]
for semantic change detection, and
\[
M_{\text{ch}} = M_{\text{ca}} \cdot M^1 + M_{\text{ca}} \cdot M^2
\]
for binary change detection [2604.11231].

CACH contains four named components: **Feature Modulation Module (FMM)**, **Bi-temporal Difference Fusion Module (BDFM)**, **Effective Difference Query Attention (EDQA)**, and **Residual Upsampler (ResUp)**. Using a **DINOv2-Base** encoder, the framework selects multi-layer features
\[
\{F_2^t, F_5^t, F_8^t, F_{11}^t\} = \Phi_{\text{DINO}}(I^t), \qquad t \in \{1,2\}.
\]
FMM aligns these features by a \(1 \times 1\) convolution and scale adjustment,
\[
\tilde{F}_i^t = \text{Resize}_i(\mathrm{Conv}_{1\times1} * F_i^t).
\]

BDFM then computes an attention map from absolute feature differences,
\[
Att_i = \sigma\left(\mathrm{Conv}_{3\times3} * \left|\tilde{F}_i^1 - \tilde{F}_i^2\right|\right),
\]
uses it to enhance each temporal feature,
\[
X_i^t = \gamma\,\mathrm{Conv}_{3\times3} * \left(\tilde{F}_i^t + Att_i \cdot \tilde{F}_i^t\right),
\]
and fuses the two streams into
\[
D_i = \gamma\,\mathrm{Conv}_{3\times3} * \left(\gamma\,\mathrm{Conv}_{3\times3} * (X_i^1 \,||\, X_i^2) \cdot Att_i\right).
\]
The paper states that this enhancement can also amplify noise and create pseudo-changes, which motivates EDQA.

EDQA calibrates these discrepancies by using the difference features \(D_i\) to query the modulated features \(\tilde{F}_i^t\). It first forms
\[
\tilde{G}_i^t = \mathrm{Conv}_{1\times1} * (D_i \,||\, \tilde{F}_i^t),
\]
then applies sliding-window attention,
\[
\tilde{D}_i^t = \phi_{\text{proj}} \left( \mathrm{Softmax}\left( \frac{\phi_{q_i}(D_i)\cdot \phi_{k_i}(D_i)^\top}{\sqrt{d_i} + b_i} \right) \cdot \phi_{v_i}^t(\tilde{G}_i^t) \right).
\]
The calibrated outputs are fused as
\[
\tilde{D}_i = \gamma\,\mathrm{Conv}_{3\times3} * \left(\gamma\,\mathrm{Conv}_{3\times3} * (\tilde{D}_i^1 || \tilde{D}_i^2)\right).
\]
Within EDQA, the MLP is augmented with a **Mixture-of-Experts (MoE)** to handle heterogeneous sensors, seasons, and imaging conditions.

ResUp progressively upsamples and aggregates refined differences,
\[
\tilde{X}_i^D = \phi_{2\times}\big(\mathrm{ResConv}_2(\mathrm{ResConv}_1(\tilde{X}_i^d) + \tilde{X}_{i-1}^D)\big),
\]
and produces the final dense category-agnostic prediction
\[
P_{\text{ca}} = \phi_{H\times W}\left(\mathrm{Conv}_{1\times1} * (\gamma\,\mathrm{Conv}_{3\times3} * \tilde{X}^D)\right), \qquad
M_{\text{ca}} = \arg\max(P_{\text{ca}}).
\]

The OVSS branch is not replaced by Seg2Change; it is wrapped by the adapter. In the main experiments the OVSS model is **SegEarth-OV3**, and text prompts include class-group formulations such as **building: \{building, roof, house\}**, **tree: \{tree, forest\}**, and **water: \{water, river\}**. The paper’s own shorthand for the interface is effectively “**segmentation for semantics + change head for change localization**.”

## 4. Training objectives and optimization regime

CACH is pre-trained on CA-CDD with a weighted combination of three losses. The first is the primary change-map loss,
\[
\mathcal{L}_{\text{cd}} = \mathcal{L}_{\text{bce}}\big(\delta_\uparrow(X_2^d, X_5^d, X_8^d, X_{11}^d), y^l\big),
\]
where \(y^l\) is the category-agnostic change label and \(\delta_\uparrow\) is the ResUp decoder. The second is a multi-layer upsampling loss,
\[
\mathcal{L}_{\text{ups}} = \sum_{i\in N_L}\mathcal{L}_{\text{bce}}\big(\delta_\uparrow(X_i^d), y^l\big).
\]
The third is a similarity loss for unchanged regions,
\[
\mathcal{L}_{\text{sim}} = \big[1 - \text{cos}(\delta_\uparrow(\tilde{F}^1), \delta_\uparrow(\tilde{F}^2))\big] \cdot \tilde{y}^l,
\]
where \(\tilde{y}^l\) denotes unchanged-region labels [2604.11231].

The total training objective is
\[
\mathcal{L}_{\text{total}} = \alpha \mathcal{L}_{\text{cd}} + \beta \mathcal{L}_{\text{ups}} + \upsilon \mathcal{L}_{\text{sim}},
\]
with \(\alpha = 0.8\), \(\beta = 0.1\), and \(\upsilon = 0.1\). The unchanged-region similarity term is specifically intended to suppress pseudo-change predictions, complementing the denoising role of EDQA.

The reported optimization configuration is: **DINOv2-Base** backbone; selected layers \(\{2,5,8,11\}\); feature channels \((48,64,80,96)\); window size \(S_w = 9\); experts \(N_e = 4\); crop size \(512 \times 512\); batch size \(4\); learning rate \(10^{-3}\); training for **20 epochs** on a **single NVIDIA RTX 4090**. These details indicate that the adapter is lightweight relative to many full end-to-end change detection systems, while still relying on a strong OVSS backbone for semantics.

## 5. Evaluation, benchmarks, and reported performance

Seg2Change is evaluated on six datasets spanning three task families. For **building change detection**, the benchmarks are **WHU-CD** and **LEVIR-CD**. For **land-cover change detection**, they are **DSIFN** and **CLCD**. For **semantic change detection**, they are **SC-SCD** and **SECOND**. The reported metrics include \(F1^c\), \(IoU^c\), Precision \(Pre^c\), Recall \(Rec^c\), Overall Accuracy, and Kappa; on semantic change detection the reported summaries are **\(mF1^c\)** and **\(mIoU^c\)** [2604.11231].

The baselines listed in the comparison are **PCA_KMeans**, **CVA**, **DCVA**, **UCD-SCM**, **AnyChange**, **Inst-CEG**, and **DynamicEarth**. Some baselines are also re-run with **SegEarth-OV3** as the identifier and marked with an asterisk for fairness. The numerical results reported for Seg2Change are:

| Benchmark group | Dataset | Reported Seg2Change result |
|---|---|---|
| Building CD | WHU-CD | 86.18 F1 / 75.72 IoU / 0.8562 Kappa |
| Building CD | LEVIR-CD | 78.72 F1 / 64.91 IoU / 0.7742 Kappa |
| Land-cover CD | DSIFN | 58.56 F1 / 41.40 IoU |
| Land-cover CD | CLCD | 47.89 F1 / 31.48 IoU |
| Semantic CD | SC-SCD | 35.68 mF1 / 23.22 mIoU / 95.82 OA / 0.3385 Kappa |
| Semantic CD | SECOND | 42.89 mF1 / 29.08 mIoU / 95.17 OA / 0.4045 Kappa |

The headline gains reported in the paper are **+9.52 IoU on WHU-CD**, **+3.87 F1 on DSIFN**, **+9.73 F1 on CLCD**, **+5.25 mIoU** and **+6.57 mF1** on **SC-SCD**, and **+5.50 mIoU** and **+5.38 mF1** on **SECOND**. The framework is also described as more efficient than **DynamicEarth** on semantic tasks, with lower GPU memory, lower inference latency, and only about **3.9M learnable parameters**.

Ablation studies attribute performance gains to each CACH component. The progression **FMM only \(\rightarrow\) +BDFM \(\rightarrow\) +EDQA \(\rightarrow\) +ResUp** is reported to improve performance on **WHU-CD** and **DSIFN**, and the best backbone feature combination is stated to be \(\{F_2, F_5, F_8, F_{11}\}\). The loss ablation further indicates that \(\mathcal{L}_{\text{sim}}\) is especially helpful for suppressing pseudo-changes, while CA-CDD pretraining gives the best results on **SECOND** compared with pretraining on the original coarse datasets.

## 6. Position within the literature, distinctions, and limitations

Seg2Change belongs to a rapidly expanding line of work that adapts segmentation or foundation-model machinery to change detection, but its target task and decomposition are distinct. It is not identical to **SEG2CD**, the conversion recipe introduced in “Exchange Is All You Need for Remote Sensing Change Detection,” where ordinary semantic segmentation models become change detectors by inserting a parameter-free exchange mechanism into a Siamese Encoder–Exchange–Decoder framework [2601.07805]. SEG2CD is centered on feature exchange and information-preserving bi-temporal fusion, whereas Seg2Change is centered on **open-vocabulary semantic segmentation models**, **category-agnostic dense change localization**, and semantic indexing for OVCD.

It is also distinct from **AnyChange**, which is a zero-shot, training-free change detection framework built on SAM via **bitemporal latent matching** and an optional **point query mechanism** [2402.01188]. AnyChange compares object proposals across time in SAM’s latent space using bidirectional negative cosine similarity, whereas Seg2Change explicitly rejects proposal-centric and threshold-driven OVCD as its primary design principle. Similarly, it differs from **SCM**, which combines FastSAM and CLIP for unsupervised building change detection through **Recalibrated Feature Fusion** and **Piecewise Semantic Attention** but is specialized to building-oriented UCD rather than open-vocabulary bi-temporal change labeling [2312.16410].

Several limitations are stated or implied in the Seg2Change paper itself. The framework still depends on the quality of the chosen OVSS model for semantic labeling. CACH is trained on re-annotated datasets derived from existing labels, so annotation coverage still reflects available data sources. The design is for **bi-temporal** change detection rather than arbitrary multi-temporal sequences. It also requires prompt or class-name mappings, although the paper states that it does not rely on heavy prompt engineering. The broader implication suggested by the work is that future improvements in OVSS backbones and larger category-agnostic change corpora could improve OVCD without redesigning the localization head [2604.11231].

Source: https://www.emergentmind.com/topics/seg2change