---
title: Open-Vocabulary Change Detection (OVCD)
url: https://www.emergentmind.com/topics/open-vocabulary-change-detection-ovcd
type: topic
---

# Open-Vocabulary Change Detection (OVCD)

Open-Vocabulary Change Detection (OVCD) is a remote-sensing change analysis task that localizes and semantically identifies changes between bi-temporal images under arbitrary user-defined text prompts rather than a fixed closed label set. In contrast to binary change detection, which asks only whether change occurred, and conventional semantic change detection, which assumes predefined categories, OVCD asks both **where** a queried concept changed and **what** semantic change occurred for concepts that may be unseen during task-specific training. The topic emerged from the convergence of change detection with vision-language and segmentation foundation models, especially CLIP-, SAM-, and DINO-family systems, and has rapidly diversified into training-free, unsupervised, weakly supervised, and adapter-based formulations for remote sensing [2501.12931][2501.06808].

## 1. Definition and emergence

DynamicEarth introduced OVCD as a task that bridges vision and language to detect changes across any category, formalizing the setting as change localization and identification between two temporally separated images \(x_{img_1}\) and \(x_{img_2}\) with arbitrary text labels \(x_{text}\) [2501.12931]. Subsequent work sharpened the formulation around remote sensing. AdaptOVCD defined a zero-shot inference mapping
\[
\mathcal{M}: (I_a, I_b, \mathcal{T}) \rightarrow M \in \{0,1\}^{H \times W},
\]
where \(I_a\) and \(I_b\) are pre- and post-change images, \(\mathcal{T}\) is the text prompt describing the target change category, and \(M\) is the binary change mask [2602.06529]. CoRegOVCD specified the queried-concept version with aligned images
\[
I_a, I_b \in \mathbb{R}^{H \times W \times 3}
\]
and output
\[
\hat{Y}^{(c)} \in \{0,1\}^{H \times W},
\]
indicating where the queried concept changes between the two dates [2604.02160].

The field developed out of dissatisfaction with closed-set change detection. Across the literature, conventional methods are described as dataset-dependent, annotation-hungry, and restricted to predefined categories, with poor generalization to unseen object types, diverse scenes, sensor geometries, and open-world user queries [2512.13089][2602.06529]. Semantic-CD represents an early transition point: it explicitly imported CLIP semantics into semantic change detection, but also noted that current SCD datasets are too small in category diversity for a fully standard OVCD benchmark, leading it to “degrade” the setting into an open-capability SCD model on SECOND [2501.06808].

A recurring definitional distinction is that OVCD is not merely open-vocabulary segmentation applied independently at two timestamps. The problem is explicitly bi-temporal and must suppress pseudo-changes caused by illumination, season, atmospheric conditions, view shifts, radiometric inconsistency, and semantic ambiguity [2604.02160][2606.20032]. This makes OVCD a joint problem of semantic querying, temporal reasoning, and robust change validation rather than a direct extension of single-image open-vocabulary recognition.

## 2. Canonical decompositions of the task

A foundational contribution of the OVCD literature is the decomposition of the task into reusable subproblems. DynamicEarth argued that a complete OVCD system should include three components—**Identifier**, **Comparator**, and **Mask Proposal**—and instantiated two training-free paradigms: **M-C-I** (*Mask Proposal → Comparator → Identifier*) and **I-M-C** (*Identifier → Mask Proposal → Comparator*) [2501.12931]. The first discovers all potential changes and then classifies them; the second identifies targets of interest first and then determines whether their states have changed.

OpenDPR further formalized OVCD as a **two-stage pipeline**: **change localization** and **category identification** [2603.27645]. Its controlled error attribution on Hi-UCD mini reported **49.1% mIoU** for **Oracle Identification**, **23.8% mIoU** for **Oracle Change Proposal**, and **13.1% mIoU** for the full **M-C-I** pipeline, leading to the diagnosis that category identification errors dominate final performance degradation. This diagnosis shifted attention away from viewing OVCD as a monolithic problem and toward separate treatment of localization and recognition.

Other papers retained the same broad decomposition but instantiated it differently. AdaptOVCD wrote the pipeline as
\[
M = \mathcal{F}_{cls}\left(\mathcal{F}_{cmp}\left(\mathcal{F}_{seg}(I_a), \mathcal{F}_{seg}(I_b)\right), \mathcal{T}\right),
\]
with \(\mathcal{F}_{seg}\) for instance segmentation, \(\mathcal{F}_{cmp}\) for feature comparison, and \(\mathcal{F}_{cls}\) for semantic identification [2602.06529]. ReA-OVCD, by contrast, treated dense pixel-wise semantic parsing as the source of candidate change regions and then introduced explicit semantic and spatial reliability checks to validate those candidates [2606.20032]. MemOVCD criticized “late-comparison systems,” writing them as
\[
\mathbf{Y}^q = c\big(h(\mathbf{T}_1,q),\, h(\mathbf{T}_2,q)\big),
\]
and reformulated OVCD as conditional cross-temporal interpretation,
\[
\mathbf{Y}^q = c\big(\phi(\mathbf{T}_1 \mid \mathbf{T}_2,q),\, \phi(\mathbf{T}_2 \mid \mathbf{T}_1,q)\big),
\]
thereby making temporal coupling itself part of the task definition [2604.26774].

These decompositions are significant because they expose different failure modes. Proposal-first pipelines emphasize candidate coverage and region comparison; identifier-first pipelines emphasize category discovery; dense posterior methods emphasize calibrated semantic discrepancy; and memory-based methods emphasize temporally conditioned semantic continuity. A plausible implication is that OVCD has evolved less as a single algorithmic lineage than as a family of competing answers to the question of where semantic change evidence should first be made reliable.

## 3. Methodological families and representative systems

The OVCD literature now contains several distinct methodological families.

| Method | Core design | Supervision mode |
|---|---|---|
| Semantic-CD | CLIP-based open semantic prompter with decoupled BCD and SCD decoders | Two-stage trained |
| DynamicEarth | Training-free M-C-I and I-M-C frameworks from off-the-shelf foundation models | Training-free |
| UniVCD | Frozen SAM2 + frozen CLIP + SCFAM + post-processing | Unsupervised |
| AdaptOVCD | ARA + ACT + ACF over SAM-HQ, DINOv3, DGTRS-CLIP | Training-free |
| OpenDPR / OpenDPR-W | Diffusion-guided prototype retrieval, optionally with S2C | Training-free / weakly supervised |
| OmniOVCD | Standalone SAM 3 with SFID | Training-free |
| CoRegOVCD | CPC + SPD + GeoGate + RCD dense posterior reasoning | Training-free |
| Seg2Change | CACH adapter plus OVSS model | Trained adapter |
| MemOVCD | Cross-temporal memory reasoning and global-local adaptive rectification | Training-free |
| ReA-OVCD | SCR + BCR over frozen SAM-3 parsing | Training-free |

One family centers on **CLIP-derived semantics**. Semantic-CD used a **bi-temporal CLIP visual encoder**, an **open semantic prompter**, a **binary change detection decoder**, and a **semantic change detection decoder**, with fully decoupled multi-task learning to separate binary localization from semantic labeling [2501.06808]. UniVCD extended this line into unsupervised OVCD by combining a **frozen SAM2 encoder**, a **frozen CLIP encoder**, a lightweight **SAM-CLIP Feature Alignment Module (SCFAM)**, and a post-processing stage, explicitly avoiding end-to-end fine-tuning of SAM2 or CLIP [2512.13089].

A second family uses **multi-model training-free pipelines**. DynamicEarth assembled off-the-shelf foundations into M-C-I and I-M-C systems [2501.12931]. AdaptOVCD organized the same general idea into **data**, **feature**, and **decision** levels, with **Adaptive Radiometric Alignment (ARA)**, **Adaptive Change Thresholding (ACT)**, and **Adaptive Confidence Filtering (ACF)** layered over **SAM-HQ**, **DINOv3**, and **DGTRS-CLIP** [2602.06529]. OpenDPR departed from text matching by building **visual prototypes** with **GPT-4**, **Diffusion-Sat**, **APE**, and **DINOv2**, then retrieving category labels in visual feature space; its optional **OpenDPR-W** added **S2C**, a weakly supervised spatial-to-change module trained with image-level binary change labels [2603.27645].

A third family pursues **single-backbone or tighter-integrated designs**. OmniOVCD argued that chaining CLIP, DINO, and SAM causes feature mismatch and instability, and instead used **SAM 3** alone with **Synergistic Fusion to Instance Decoupling (SFID)** [2601.13895]. CoRegOVCD likewise favored dense inference over explicit multi-model association, converting raw concept responses into competition-aware posteriors and then regularizing them geometrically and regionally [2604.02160]. Seg2Change adapted an open-vocabulary semantic segmentation model to change detection by coupling a trained **category-agnostic change head (CACH)** with an OVSS backbone, specifically **SegEarth-OV3** in the main experiments [2604.11231].

A fourth family emphasizes **temporal coupling and reliability correction**. MemOVCD reformulated bi-temporal change detection as a **two-frame tracking problem** using **SAM 3.1**, **weighted bidirectional propagation**, **histogram-aligned transition frames**, and **global-local adaptive rectification** [2604.26774]. ReA-OVCD began from pixel-wise semantic discrepancies produced by a frozen **SAM-3** parser and then validated them with **Semantic Change Reasoning (SCR)** and **Boundary-aware Change Refinement (BCR)** [2606.20032].

## 4. Core algorithmic mechanisms

Despite architectural diversity, several recurring mechanisms organize the technical core of OVCD.

**Proposal-based comparison** is prominent in early training-free systems. DynamicEarth’s ovcdred computes a mask-level change score with latent feature matching,
\[
\mathcal{D}(\mathbf{m}) = -\frac{\mathbf{z_1[\mathbf{m}]}}{\|\mathbf{z_1[\mathbf{m}]}\|_2} \cdot \frac{\mathbf{z_2[\mathbf{m}]}}{\|\mathbf{z_2[\mathbf{m}]}\|_2},
\]
after generating class-agnostic masks and applying NMS [2501.12931]. OpenDPR retains the proposal-then-identify logic but replaces text matching with prototype retrieval in visual space. For a proposal \(z_i\), it evaluates
\[
\hat{c}_i = \arg\max_c \max_k \text{sim}(z_i, P_{c,k}),
\]
and reports that **global-max usually performs better** than category-mean because it preserves strong semantic matches and avoids averaging away discriminative sub-attributes [2603.27645].

**Cross-modal semantic alignment** defines another major line. UniVCD’s SCFAM aligns SAM2’s spatial detail with CLIP’s semantic priors through a layer-wise fusion block, lightweight adapters, and multiple projection heads. Change is then computed from per-image class-probability maps. For category \(c\),
\[
D_c(i,j)=\bigl(S_{1,c}(i,j)-S_{2,c}(i,j)\bigr)^2,
\]
making change estimation category-wise rather than purely binary [2512.13089]. This category-wise formulation is what the paper identifies as the mechanism enabling a unified treatment of binary, semantic, and multi-class change detection.

**Dense posterior discrepancy** replaces explicit instance matching in CoRegOVCD. After **Competitive Posterior Calibration (CPC)**, it defines the **Semantic Posterior Delta (SPD)** as
\[
\Delta^{(c)}(x)=\left| P_a^{(c)}(x) - P_b^{(c)}(x) \right|.
\]
The discrepancy is then filtered by **Geometry-Token Consistency Gate (GeoGate)** using **Depth Anything 3**, and regionally regularized by **Regional Consensus Discrepancy (RCD)** with SLIC superpixels [2604.02160]. This shifts change inference from discrete object correspondence to comparison of competition-aware queried-concept posteriors.

**Pixel-wise parsing plus reliability reasoning** is the logic of ReA-OVCD. It begins with hard label discrepancy,
\[
H_{\text{change}}(x)=I\big(Y_1(x)\neq Y_2(x)\big),
\]
then evaluates semantic reliability through Jensen-Shannon divergence and maximum per-class logit difference, and finally validates candidate regions spatially by measuring whether they contain reliable interior pixels far from unstable boundaries [2606.20032]. The paper explicitly positions this as a response to the trade-off between coarse but stable instance matching and detailed but noisy direct pixel comparison.

**Temporal propagation and memory reasoning** are most explicit in MemOVCD. Coarse query-specific masks initialize SAM 3 memory, mask evidence is propagated bidirectionally over **histogram-aligned transition frames**, stable regions are pooled into a query-specific visual exemplar, and patch-wise and global-view predictions are fused by connected-component-aware global-local rectification [2604.26774]. This is one of the clearest departures from OVCD as static differencing.

**Change-specific adapters** appear in Seg2Change. Its **Feature Modulation Module (FMM)**, **Bi-temporal Difference Fusion Module (BDFM)**, **Effective Difference Query Attention (EDQA)**, and **Residual Upsampler (ResUp)** learn a **category-agnostic change map** from CA-CDD, after which semantic segmentation maps index changed pixels to classes [2604.11231]. This avoids proposal generation and fixed similarity thresholds.

## 5. Benchmarks and empirical landscape

OVCD is evaluated on a recurring set of public benchmarks, but protocols vary substantially across papers.

| Benchmark | Typical use in OVCD papers | Notes appearing in the literature |
|---|---|---|
| LEVIR-CD | Building change detection | 637 pairs total in UniVCD; 128 test pairs |
| WHU-CD / WHU-CD-256 | Building change detection | Patch and original-image evaluation both appear |
| SECOND | Semantic or multi-class change detection | Six classes: building, tree, water, low vegetation, non-vegetated surface, playground |
| DSIFN | Land-cover or building-related change | Used in AdaptOVCD, Seg2Change, ReA-OVCD |
| S2Looking | Building change detection | Used in DynamicEarth, OmniOVCD, MemOVCD |
| BANDON | Building change detection | Used in DynamicEarth and MemOVCD |
| Hi-UCD mini | Error attribution / semantic change detection | Used in OpenDPR |
| SC-SCD / CLCD | Semantic or land-cover change detection | Used in Seg2Change |

DynamicEarth established an initial training-free baseline regime. Its best building result in the reported table was **SAM-DINOv2-SegEarth-OV** with **36.6 IoU / 53.6 F1** on LEVIR-CD and **40.6 IoU / 57.7 F1** on WHU-CD; it also reported that OVCD methods beat the best cross-dataset supervised result on all test sets by **+7.9 IoU / +7.0 F1** on LEVIR-CD, **+10.8 IoU / +8.9 F1** on WHU-CD, **+11.9 IoU / +17.1 F1** on S2Looking, and **+13.8 IoU / +22.8 F1** on BANDON [2501.12931].

Later methods improved markedly but under different settings. UniVCD reported **UniVCD (orig., postproc)** with **F1 = 70.7**, **IoU = 54.7**, **mIoU = 75.6** on LEVIR-CD and **F1 = 76.5**, **IoU = 61.9**, **mIoU = 79.6** on WHU-CD, while highlighting strong category-wise performance on SECOND, especially **building: F1 58.4, IoU 41.2** and **non-vegetated surface: F1 42.6, IoU 27.1** [2512.13089]. AdaptOVCD reported \(F_1^C\) values of **68.00%** on LEVIR-CD, **76.53%** on WHU-CD, **59.47%** on DSIFN, and **63.81%** on Building on SECOND, and stated that it reaches **84.89% of the fully-supervised performance upper bound** in cross-dataset evaluation [2602.06529].

OmniOVCD reported **67.2 IoU / 80.4 F1** on LEVIR-CD, **66.5 IoU / 79.9 F1** on WHU-CD, **24.5 IoU / 39.4 F1** on S2Looking, and a SECOND class average of **27.1 / 41.8** [2601.13895]. CoRegOVCD emphasized relative gains over the strongest previous training-free baseline, reporting **+2.61 F1\(_C\)** on LEVIR-CD, **+2.24 F1\(_C\)** on WHU-CD-256, **+4.98 F1\(_C\)** on DSIFN, and **+1.88 F1\(_C\)** on SECOND-Building, with a six-class SECOND average of **47.50% F1\(_C\)** and **31.67% IoU\(_C\)** [2604.02160]. MemOVCD reported **72.5 \(IoU^c\)** and **84.1 \(F_1^c\)** on LEVIR-CD, **26.0 \(IoU^c\)** and **41.3 \(F_1^c\)** on S2Looking, **23.2 \(IoU^c\)** and **37.7 \(F_1^c\)** on BANDON, and **37.8 \(IoU^c\)** and **54.9 \(F_1^c\)** on DSIFN [2604.26774].

Other strands reached strong results under their own protocols. Seg2Change reported **WHU-CD: \(F1^c = 86.18\), \(IoU^c = 75.72\)**, **LEVIR-CD: \(F1^c = 78.72\), \(IoU^c = 64.91\)**, and **SECOND: \(mF1^c = 42.89\), \(mIoU^c = 29.08\)** [2604.11231]. ReA-OVCD reported **LEVIR-CD: \(F1_C = 82.53\), \(IoU_C = 70.26\)**, **WHU-CD: \(F1_C = 90.31\), \(IoU_C = 82.33\)**, **DSIFN: \(F1_C = 67.79\), \(IoU_C = 51.27\)**, and a SECOND class-average of **47.19% \(F1_C\)** and **32.02% IoU\(_C\)** [2606.20032]. OpenDPR, focused on the recognition bottleneck, reported **37.3 mIoU / 54.3 mF1** in SECOND multi-class inference and **OpenDPR-W** average **mIoU 19.7** and **mF1 30.4** on Hi-UCD mini [2603.27645].

Because some methods are training-free, some are unsupervised, some are weakly supervised, and some train adapters or heads, direct numerical comparison across papers is not always protocol-equivalent. This suggests that the empirical landscape is best read as evidence of rapid progress under multiple supervision and inference regimes rather than as a single strictly ordered leaderboard.

## 6. Limitations, controversies, and research directions

Several limitations recur across the OVCD literature. The first is **dataset and benchmark scarcity**. Semantic-CD explicitly states that available SCD datasets are too small in category diversity for a fully standard OVCD benchmark [2501.06808], while DynamicEarth emphasizes that existing change detection datasets are limited, scattered, and heterogeneous, with poor annotation quality and no standardized open-vocabulary labels [2501.12931]. Seg2Change’s construction of **CA-CDD** can be read as a response to this gap, but it also indicates that broader category-agnostic supervision remains a bottleneck [2604.11231].

The second is disagreement over the primary technical bottleneck. OpenDPR argues that **category identification** is the main bottleneck, not localization, because CLIP-style image-text matching compresses fine-grained land-cover semantics too aggressively for remote sensing [2603.27645]. ReA-OVCD instead frames the problem as a trade-off between **instance-level comparison**, which misses fine-grained or partial changes, and **direct pixel comparison**, which is unstable and boundary-noisy [2606.20032]. MemOVCD criticizes weak temporal coupling in late-comparison systems [2604.26774]. These are not mutually exclusive diagnoses, but they emphasize different points of failure.

The third limitation is **domain sensitivity**. Multiple papers note that remote sensing differs from natural-image pretraining in viewpoint, scale variation, orientation ambiguity, radiometric variation, and land-cover subtlety [2602.06529][2603.27645]. Concrete manifestations are category-specific failures. UniVCD reports weaker performance on water with standard CLIP and notes that replacing CLIP with **RemoteCLIP** improves the water category substantially [2512.13089]. AdaptOVCD lists **Tree** and **Low Vegetation** as harder classes due to semantic ambiguity and weak boundaries [2602.06529]. ReA-OVCD notes weaker performance on **Playground** because of sample scarcity and semantic ambiguity [2606.20032].

The fourth limitation is **post-processing and threshold sensitivity**. UniVCD’s post-processing can improve precision and IoU but may reduce recall in ambiguous scenes [2512.13089]. AdaptOVCD states that prompts and thresholds need scene-specific tuning [2602.06529]. MemOVCD reports that moderate histogram-aligned bridging helps substantially, with \(K=3\) performing best, while \(K=5\) begins to degrade [2604.26774]. CoRegOVCD and ReA-OVCD both depend on calibrated dense scores and structural filters, underscoring that fully training-free inference often shifts complexity from optimization to calibration and rule design [2604.02160][2606.20032].

Research directions identified in the papers are comparatively consistent. They include stronger **remote sensing-adapted foundation models**, improved **prompt engineering** or **LLM-generated prompt vocabularies**, better handling of **multi-class inference**, more reliable treatment of **small objects** and **internal-part changes**, and more diverse datasets with richer open-vocabulary annotations [2501.12931][2602.06529]. Another clear trajectory is toward greater temporal coupling and internal model coherence: OmniOVCD moves to a standalone SAM 3 design [2601.13895], MemOVCD turns OVCD into memory-based tracking [2604.26774], and CoRegOVCD and ReA-OVCD replace brittle direct differencing with calibrated posterior or reliability-aware reasoning [2604.02160][2606.20032]. This suggests a broader consolidation of OVCD around semantically calibrated, temporally aware, and structurally validated inference rather than around naïve before/after comparison alone.

Source: https://www.emergentmind.com/topics/open-vocabulary-change-detection-ovcd