Pretrained vision-language models such as CLIP excel at zero-shot recognition but often fail at compositionality, particularly attribute-object and relational structures. Recent studies mitigate this issue by augmenting training with synthetic hard negatives generated by a cascade of large language models and text-to-image models, which incurs substantial pipeline overhead. We instead propose MLLMCLIP, a heterogeneous distillation framework that transfers multimodal knowledge directly from a generative Multimodal Large Language Model (MLLM) teacher into a discriminative CLIP student, bypassing synthetic data entirely. To bridge the architectural mismatch between the two paradigms, we introduce an attention-based per-layer token selection and a CKA-based distillation loss. Compared to prior CLIP-enhancement methods, MLLMCLIP achieves state-of-the-art compositional accuracy while delivering consistent gains on standard zero-shot classification and image-text retrieval, showing that feature-level distillation strengthens both compositional and general vision-language representation capability.
Text-only training is a popular paradigm in zero-shot video captioning, where the video distribution is not available to the model during training, leading to a cross-modal gap between the training (text-only) and the inference (video-only). Previous works attempt to bridge the gap through simple linear transformations. However, the inherent gap between text and video makes cross-modal representation space alignment insufficient, resulting in inaccurate sentences. To address this issue, we propose a novel zero-shot video captioning framework (WSV) consisting of two training stages, which first generates corresponding synthetic video latent representations via a pretrained text-to-video generation model. To strengthen the fidelity of the latent representations, we propose a polisher capable of bridging the gap between real and synthetic video distributions. Subsequently, we design a prompter that conditions GPT-2 on the polished latent representations to generate the captions in the second training stage. During inference, an input video is encoded by a pretrained 3D Causal VAE and then fed directly into the prompter, which in turn guides GPT-2 to produce the final caption. Experimental results conducted on MSVD, MSR-VTT, and VATEX datasets demonstrate that our proposed method achieves scores of 52 and 95.7 on the B@4 and CIDEr metrics, respectively.
Long Hoang Dang, Tuan Nguyen Huu, Nguyen Minh Hieu +1cs.CV
Adapting CLIP for zero-shot sketch-based image retrieval (ZS-SBIR) via prompt learning faces a fundamental tension: the model must bridge the sketch-photo domain gap through task-specific adaptation, yet the added flexibility risks overfitting to seen training categories and eroding CLIP's zero-shot generalization. We present SeCo-SBIR, a semantically consistent prompt learning framework that resolves this tension from both sides. First, a text-guided multi-modal prompting strategy routes learnable prompt vectors through CLIP's text encoder and projects the resulting intermediate representations into the visual encoder at every layer via learnable coupling functions. Because the text encoder has already learned robust, abstract category-level semantics from large-scale language supervision, this mechanism injects transferable semantic knowledge directly into the visual pathway - adapting the model to the sketch-photo domain while inherently favoring generalization to unseen classes. Second, a perturbation-based consistency constraint addresses the residual overfitting risk from the learnable coupling functions by aligning the adapted model with a frozen CLIP reference branch using an asymmetric InfoNCE objective - augmented inputs feed the frozen branch while clean inputs feed the trainable branch - anchoring the learned representations to CLIP's generalizable feature space. Together with lightweight adapters and a multi-objective loss combining triplet, NT-Xent, and classification terms, SeCo-SBIR achieves state-of-the-art results on all three standard ZS-SBIR benchmarks across categorical, generalized, and across-dataset settings.
Zero-shot image captioning aims to generate image descriptions without annotated image-text pairs. Recent approaches exploit text-to-image models to synthesize training data from text-only corpora, but most focus on improving overall data quality. In contrast, we observe that synthetic image-text misalignment is often structured and fine-grained: pairs may remain globally plausible while containing missing entities or misgrounded attributes, thereby degrading supervision fidelity. As a result, methods based on global similarity for image rematching or regeneration may improve apparent plausibility, but cannot systematically repair entity-level misalignment. To address this issue, we propose ReCap, a plug-and-play framework that shifts synthetic data refinement from implicit global matching to explicit fine-grained realignment. Specifically, ReCap enforces entity-level correspondence by using detected image-supported entities to guide caption rewriting, yielding more faithful synthetic supervision. In addition, we introduce an adaptive dynamic weighted learning strategy to downweight unreliable synthetic pairs during training. As a general framework, ReCap can be integrated into existing synthetic-data pipelines. Extensive experiments show that ReCap consistently improves image-text consistency and achieves state-of-the-art performance on both in-domain and cross-domain zero-shot image captioning benchmarks.
Zero-shot composed image retrieval (ZS-CIR) aims to retrieve a target image from a multimodal query consisting of a reference image and an edit text describing the desired modification. Recent ZS-CIR studies have relied on projection-based methods that map a reference image into pseudo-word tokens in the text embedding space. However, such methods require additional projection and re-encoding steps, increasing training complexity, reducing efficiency, and introducing a discrepancy between training and inference. In this paper, we propose DiCE-CIR, a direct composition learning method that predicts composed query representations by directly composing a reference image and an edit text. To enable scalable training without manually annotated triplets, we automatically construct compositional training samples from large-scale image-caption pairs using a large language model. Based on these samples, we train a lightweight composition module with objectives that promote alignment with the target, edit-consistent semantic transformation, and retrieval discriminability. We conduct extensive experiments on ZS-CIR benchmarks and show that DiCE-CIR achieves state-of-the-art performance on CIRCO and competitive performance on CIRR while maintaining high computational efficiency.
Zero-shot composed image retrieval (ZS-CIR) aims to retrieve a target image by editing a reference image with a natural-language instruction, without relying on domain-specific annotated triplets. Most existing ZS-CIR methods rely on textual inversion to translate the reference image into pseudo-text tokens and then compose them with the instruction via simple concatenation in the text space, which can be lossy and brittle for fine-grained semantics. In this work, we propose a new paradigm, namely FlowCIR, that casts ZS-CIR as conditional semantic transport between reference and target embeddings. Leveraging \emph{conditional flow matching}, our model learns a lightweight transport field that maps the instruction representation toward a target-aligned query embedding conditioned on the reference image. Since FlowCIR operates on pre-extracted VLM embeddings and trains only a small transport module without updating the image or text encoder, it offers a computationally efficient training protocol compared with prior textual-inversion-based approaches. The resulting framework is training-efficient, requiring roughly $10\times$ fewer training resources than prior textual-inversion-based approaches. We further identify negation and removal as a major failure mode of VLM-based composition. To address this, we propose an inference-only Multi-Negative Steering strategy that steers a negation-containing relative instruction away from its negated semantics, mitigating the limited negation handling of VLMs and improving robustness on negation-heavy queries. Extensive experiments on standard CIR benchmarks demonstrate that FlowCIR achieves strong and competitive performance compared with recent ZS-CIR methods.
Zero-shot video temporal grounding (VTG) localizes events in untrimmed videos from natural language queries without task-specific training. Existing methods rely on frame-query feature matching, which suffices for simple events but struggles with complex multi-stage queries that require understanding temporal ordering and causal structure -- a disparity we call the reasoning gap. We propose DART (Difficulty-Adaptive Routing for Temporal Grounding), which bridges this gap by coupling difficulty-aware routing with structured reasoning in large vision-language models. A query-conditioned Determinantal Point Process (DPP) serves a dual role: selecting diverse, query-relevant keyframes as temporal evidence, and providing spectral entropy as a difficulty indicator. Simple queries are routed to a Fast path for direct prediction, while complex queries follow a Slow path with Temporal Markup Prompting, which decomposes localization into global event analysis, per-frame temporal role annotation, and boundary extraction. On Charades-STA and ActivityNet Captions, DART achieves state-of-the-art zero-shot performance across both identically distributed and multiple out-of-distribution settings, improving mIoU by up to 3.5 points over the strongest baseline while using over 7 times fewer frames. The project homepage is available at https://dart-vtg.github.io/.
Jingjing Zhang, Lei Zhang, Zheren Fu +1cs.CV cs.AI cs.CL cs.IR cs.MM
Composed Image Retrieval (CIR) retrieves a target image from a reference image and a textual modification. While supervised CIR relies on costly triplets, Zero-Shot CIR (ZS-CIR) alleviates this reliance through proxy tasks trained on image-text pairs. However, existing proxy tasks primarily enhance visual and textual representations to accommodate a predefined composition mechanism such as pseudo-word injection into a frozen text encoder or linear feature arithmetic. As a result, the composition function itself remains unlearned, limiting the model's ability to express diverse and fine-grained semantic modifications. To address this, we propose FoCo, which models composition as two coordinated stages: focusing on modification-relevant visual content, and then completing the target semantics. We realize these through two proxy tasks: text-anchored visual aggregation to selectively gather visual content guided by localized textual semantics, and context-conditioned semantic completion to transform these aggregated visuals with the remaining scene context into a coherent composed representation. The tasks are trained jointly with a cross-instance contrastive objective, encouraging semantic diversity and discouraging shortcut composition strategies. Extensive experiments on four ZS-CIR benchmarks show FoCo's state-of-the-art performance and improved generalization.
Due to the increasing frequency and intensity of extreme climate events, there is a clear demand for intelligent, scalable, and autonomous approaches to disaster damage assessment. Existing methods, largely based on supervised learning and task-specific fine-tuning, struggle to generalize under domain shifts, long-tailed data distributions, and heterogeneous geospatial data sources, especially in disaster scenarios. They also often lack the ability to integrate and reason across multimodal geospatial information, such as satellite images and street-view images. In this paper, we introduce RAPID, a reproducible multi-agent pipeline for interpretable disaster damage assessment, including damage-level assessment, damage-type interpretation, and actionable suggestions for response, remediation, and recovery. RAPID coordinates specialized agents to perform cross-view understanding, image restoration, structured damage recognition, and geographical reasoning across heterogeneous data modalities. Without task-specific fine-tuning, RAPID supports zero-shot damage assessment by jointly using complementary information from remote sensing and ground-level perspectives. The system produces fine-grained, interpretable assessments and automatically generates location-specific, decision-relevant disaster reports to support early-stage emergency response. We evaluate RAPID across hurricanes, floods, wildfires, and earthquakes using multiple cross-view imagery inputs, including pre- and post-disaster street-view images, post-disaster remote sensing imagery, and street-view image pairs. Experiments show that RAPID achieves 0.92 overall accuracy for multi-disaster type classification and up to 0.627 for cross-view damage severity prediction, highlighting its potential as a foundational framework for autonomous disaster intelligence.
This paper presents Custom ZeroCLIP, a retrieval-augmented vision-language framework for zero-shot captioning of Indonesian traditional garments. The dataset contains 3,800 expert-annotated images from all 38 Indonesian provinces. Using a province-level inductive zero-shot protocol, the model is trained on 24 seen provinces, validated on 6 seen provinces, and evaluated on 8 unseen provinces. The framework combines a frozen CLIP ViT-B/32 image encoder, a CLIP text encoder, a BERT text encoder, and an LSTM caption decoder. During inference, unseen-province labels and captions are unavailable, and retrieval uses only captions from training provinces. No unseen-province image, label, or caption is used during training, validation, or retrieval-bank construction. Custom ZeroCLIP achieves a CLIPScore of 0.8536, BLEU-4 of 0.3342, and METEOR of 0.4859, outperforming existing baselines. Ablation results show that retrieval improves cultural vocabulary recovery with a 19.3\% METEOR gain, while human evaluation confirms stronger cultural accuracy and fluency. The results demonstrate the effectiveness of retrieval-augmented domain adaptation for culturally grounded caption generation in low-resource heritage settings. The dataset is publicly available at https://github.com/AnugrahAidinYotolembah/Traditional-Indonesian-Clothing-Captioning-Dataset.
Audio-visual Generalized Zero-shot Learning (AV-GZSL) is a challenging task that aims to classify both seen and unseen objects or scenes by integrating data from audio and visual modalities. Recent studies primarily focus on fusing or aligning audio and visual features to generate more informative audio-visual embeddings. Also, aligning the audio-visual and textual features of most existing methods relies solely on the optimization objectives. However, those methods neglect the inherent distributional and structural differences between audio-visual and textual modalities. To address this limitation, we propose a method termed Aligning Hierarchical Standardized Embedding (AHSE), which enables hierarchical alignment of standardized audio-visual and textual embeddings within a shared embedding space. Specifically, we first apply Z-score standardization to the fused audio-visual and textual embeddings to reduce distributional mismatches. We then introduce a hierarchical alignment strategy that minimizes discrepancies at the semantic, class, and batch levels, thereby constructing a more robust and well-structured embedding space. This strategy not only preserves semantic and inter-class relationships but also maintains spatial consistency within each batch. Extensive experiments on three benchmark datasets: VGGSound-GZSL, UCF-GZSL, and ActivityNet-GZSL, demonstrate that AHSE achieves competitive performance in zero-shot learning.
Zero-shot learning (ZSL) for inertial measurement unit (IMU)-based human activity recognition (HAR) faces a central challenge: bridging the gap between sensor embeddings and semantic class representations. We systematically evaluate seven configurations combining three inference methods with two training pipelines on the PAMAP2 dataset, using 14 seen and 4 unseen activity classes with subjects 108 and 109 held out for testing. We find that the modality gap is a training-time phenomenon governed by the encoder objective. A temporal convolutional network (TCN) trained with cross-entropy over label-name Sentence- BERT prototypes yields sensor embeddings with a mean cosine similarity of 0.30 to the corresponding text prototypes, while replacing the label-name prototype targets with discriminative activity descriptions raises this to 0.69. This alignment improvement transfers consistently across all three inference methods. The strongest result combines contrastive training with inverted softmax correction, achieving 73.2% accuracy and 0.583 macro F1 on unseen classes, compared to 58.3% accuracy and 0.34 macro F1 for the label-name baseline. A secondary finding is that richer text descriptions reduce inter-prototype separability in Sentence-BERT space, because shared biomechanical vocabulary causes the language model to compress the prototype cloud. This effect does not negate the benefits of contrastive alignment provided prototype descriptions retain sufficient discriminative vocabulary. We also demonstrate that overall accuracy is a misleading primary metric when test-set class distributions are imbalanced, and recommend macro-averaged F1 as the standard reporting metric for ZSL-HAR benchmarks.
Zero-shot 3D Visual Grounding (3DVG) is a critical capability for open-world embodied AI. However, existing methods are fundamentally bottlenecked by the poor quality of open-vocabulary 3D proposals, suffering from inaccurate categories and imprecise geometries, as well as the spatial redundancy of exhaustive multi-view reasoning. To address these challenges, we propose MCM-VG, a novel framework that achieves robust zero-shot 3DVG by explicitly establishing Multiple Consistent 2D-3D Mappings. Instead of passively relying on noisy 3D segments, MCM-VG enforces 2D-3D consistency across three fundamental dimensions to achieve precise target localization and reliable reasoning. First, a Semantic Alignment module corrects category mismatches via LLM-driven query parsing and coarse-to-fine 2D-3D matching. Second, an Instance Rectification module leverages VLM-guided 2D segmentations to reconstruct missing targets, back-projecting these reliable visual priors to establish accurate 3D geometries. Finally, to eliminate spatial redundancy, a Viewpoint Distillation module clusters 3D camera directions to extract optimal frames. By pairing these optimal RGB frames with Bird's Eye View maps into concise visual prompt sets, we formulate the final target disambiguation as a multiple-choice reasoning task for Vision-Language Models. Extensive evaluations on ScanRefer and Nr3D benchmarks demonstrate that MCM-VG sets a new state-of-the-art for zero-shot 3D visual grounding. Remarkably, it achieves 62.0\% and 53.6\% in Acc@0.25 and Acc@0.5 on ScanRefer, outperforming previous baselines by substantial margins of 6.4\% and 4.0\%.