Pixel-level annotation is expensive and time-consuming. Semi-supervised segmentation methods address this challenge by learning models on few labeled images alongside a large corpus of unlabeled images. Although foundation models could further account for label scarcity, effective mechanisms for their exploitation remain underexplored. We address this by devising a novel semi-supervised panoptic approach fueled by two dedicated foundation models. We enhance recognition by complementing unsupervised mask-transformer consistency with zero-shot classification of CLIP features. We enhance localization by class-agnostic decoder warm-up with respect to SAM pseudo-labels. The resulting decoupled enhancement of recognition and localization (DEARLi) particularly excels in the most challenging semi-supervised scenarios with large taxonomies and limited labeled data. Moreover, DEARLi outperforms the state of the art in semi-supervised semantic segmentation by a large margin while requiring 8x less GPU memory, in spite of being trained only for the panoptic objective. We observe 29.9 PQ and 38.9 mIoU on ADE20K with only 158 labeled images. The source code is available at https://github.com/helen1c/DEARLi.
Mammographic breast density classification is essential for cancer risk assessment but remains challenging due to subjective interpretation and inter-observer variability. This study compares multimodal and CNN-based methods for automated classification using the BI-RADS system, evaluating BioMedCLIP and ConvNeXt across three learning scenarios: zero-shot classification, linear probing with textual descriptions, and fine-tuning with numerical labels. Results show that zero-shot classification achieved modest performance, while the fine-tuned ConvNeXt model outperformed the BioMedCLIP linear probe. Although linear probing demonstrated potential with pretrained embeddings, it was less effective than full fine-tuning. These findings suggest that despite the promise of multimodal learning, CNN-based models with end-to-end fine-tuning provide stronger performance for specialized medical imaging. The study underscores the need for more detailed textual representations and domain-specific adaptations in future radiology applications.
CLIP-based domain generalization aims to improve model generalization to unseen domains by leveraging the powerful zero-shot classification capabilities of CLIP and multiple source datasets. Existing methods typically train a single model across multiple source domains to capture domain-shared information. However, this paradigm inherently suffers from two types of conflicts: 1) sample conflicts, arising from noisy samples and extreme domain shifts among sources; and 2) optimization conflicts, stemming from competition and trade-offs during multi-source training. Both hinder the generalization and lead to suboptimal solutions. Recent studies have shown that model merging can effectively mitigate the competition of multi-objective optimization and improve generalization performance. Inspired by these findings, we propose Harmonizing and Merging (HAM), a novel source model merging framework for CLIP-based domain generalization. During the training process of the source models, HAM enriches the source samples without conflicting samples, and harmonizes the update directions of all models. Then, a redundancy-aware historical model merging method is introduced to effectively integrate knowledge across all source models. HAM comprehensively consolidates source domain information while enabling mutual enhancement among source models, ultimately yielding a final model with optimal generalization capabilities. Extensive experiments on five widely used benchmark datasets demonstrate the effectiveness of our approach, achieving state-of-the-art performance.
Medical vision-language alignment through cross-modal contrastive learning shows promising performance in image-text matching tasks, such as retrieval and zero-shot classification. However, conventional cross-modal contrastive learning (CLIP-based) methods suffer from suboptimal visual representation capabilities, which also limits their effectiveness in vision-language alignment. In contrast, although the models pretrained via multimodal masked modeling struggle with direct cross-modal matching, they excel in visual representation. To address this contradiction, we propose ALTA (ALign Through Adapting), an efficient medical vision-language alignment method that utilizes only about 8% of the trainable parameters and less than 1/5 of the computational consumption required for masked record modeling. ALTA achieves superior performance in vision-language matching tasks like retrieval and zero-shot classification by adapting the pretrained vision model from masked record modeling. Additionally, we integrate temporal-multiview radiograph inputs to enhance the information consistency between radiographs and their corresponding descriptions in reports, further improving the vision-language alignment. Experimental evaluations show that ALTA outperforms the best-performing counterpart by over 4% absolute points in text-to-image accuracy and approximately 6% absolute points in image-to-text retrieval accuracy. The adaptation of vision-language models during efficient alignment also promotes better vision and language understanding. Code is publicly available at https://github.com/DopamineLcy/ALTA.
Classifying geospatial imagery remains a major bottleneck for applications such as disaster response and land-use monitoring-particularly in regions where annotated data is scarce or unavailable. Existing tools (e.g., RS-CLIP) that claim zero-shot classification capabilities for satellite imagery nonetheless rely on task-specific pretraining and adaptation to reach competitive performance. We introduce GeoVision Labeler (GVL), a strictly zero-shot classification framework: a vision Large Language Model (vLLM) generates rich, human-readable image descriptions, which are then mapped to user-defined classes by a conventional Large Language Model (LLM). This modular, and interpretable pipeline enables flexible image classification for a large range of use cases. We evaluated GVL across three benchmarks-SpaceNet v7, UC Merced, and RESISC45. It achieves up to 93.2% zero-shot accuracy on the binary Buildings vs. No Buildings task on SpaceNet v7. For complex multi-class classification tasks (UC Merced, RESISC45), we implemented a recursive LLM-driven clustering to form meta-classes at successive depths, followed by hierarchical classification-first resolving coarse groups, then finer distinctions-to deliver competitive zero-shot performance. GVL is open-sourced at https://github.com/microsoft/geo-vision-labeler to catalyze adoption in real-world geospatial workflows.
We present Distill CLIP (DCLIP), a fine-tuned variant of the CLIP model that enhances multimodal image-text retrieval while preserving the original model's strong zero-shot classification capabilities. CLIP models are typically constrained by fixed image resolutions and limited context, which can hinder their effectiveness in retrieval tasks that require fine-grained cross-modal understanding. DCLIP addresses these challenges through a meta teacher-student distillation framework, where a cross-modal transformer teacher is fine-tuned to produce enriched embeddings via bidirectional cross-attention between YOLO-extracted image regions and corresponding textual spans. These semantically and spatially aligned global representations guide the training of a lightweight student model using a hybrid loss that combines contrastive learning and cosine similarity objectives. Despite being trained on only ~67,500 samples curated from MSCOCO, Flickr30k, and Conceptual Captions-just a fraction of CLIP's original dataset-DCLIP significantly improves image-text retrieval metrics (Recall@K, MAP), while retaining approximately 94% of CLIP's zero-shot classification performance. These results demonstrate that DCLIP effectively mitigates the trade-off between task specialization and generalization, offering a resource-efficient, domain-adaptive, and detail-sensitive solution for advanced vision-language tasks. Code available at https://anonymous.4open.science/r/DCLIP-B772/README.md.
Contrastive Language-Image Pretraining (CLIP) has demonstrated strong zero-shot performance across diverse downstream text-image tasks. Existing CLIP methods typically optimize a contrastive objective using negative samples drawn from each minibatch. To achieve robust representation learning, these methods require extremely large batch sizes and escalate computational demands to hundreds or even thousands of GPUs. Prior approaches to mitigate this issue often compromise downstream performance, prolong training duration, or face scalability challenges with very large datasets. To overcome these limitations, we propose AmorLIP, an efficient CLIP pretraining framework that amortizes expensive computations involved in contrastive learning through lightweight neural networks, which substantially improves training efficiency and performance. Leveraging insights from a spectral factorization of energy-based models, we introduce novel amortization objectives along with practical techniques to improve training stability. Extensive experiments across 38 downstream tasks demonstrate the superior zero-shot classification and retrieval capabilities of AmorLIP, consistently outperforming standard CLIP baselines with substantial relative improvements of up to 12.24%.
We introduce Perceptual-Initialization (PI), a paradigm shift in visual representation learning that incorporates human perceptual structure during the initialization phase rather than as a downstream fine-tuning step. By integrating human-derived triplet embeddings from the NIGHTS dataset to initialize a CLIP vision encoder, followed by self-supervised learning on YFCC15M, our approach demonstrates significant zero-shot performance improvements, without any task-specific fine-tuning, across 29 zero shot classification and 2 retrieval benchmarks. On ImageNet-1K, zero-shot gains emerge after approximately 15 epochs of pretraining. Benefits are observed across datasets of various scales, with improvements manifesting at different stages of the pretraining process depending on dataset characteristics. Our approach consistently enhances zero-shot top-1 accuracy, top-5 accuracy, and retrieval recall (e.g., R@1, R@5) across these diverse evaluation tasks, without requiring any adaptation to target domains. These findings challenge the conventional wisdom of using human-perceptual data primarily for fine-tuning and demonstrate that embedding human perceptual structure during early representation learning yields more capable and vision-language aligned systems that generalize immediately to unseen tasks. Our work shows that "beginning with you", starting with human perception, provides a stronger foundation for general-purpose vision-language intelligence.
Pretrained vision-language models (VLMs), e.g., CLIP, demonstrate impressive zero-shot capabilities on downstream tasks. Prior research highlights the crucial role of visual augmentation techniques, like random cropping, in alignment with fine-grained class descriptions generated by large language models (LLMs), significantly enhancing zero-shot performance by incorporating multi-view information. However, the inherent randomness of these augmentations can inevitably introduce background artifacts and cause models to overly focus on local details, compromising global semantic understanding. To address these issues, we propose an \textbf{A}ttention-\textbf{B}ased \textbf{S}election (\textbf{ABS}) method from local details to global context, which applies attention-guided cropping in both raw images and feature space, supplement global semantic information through strategic feature selection. Additionally, we introduce a soft matching technique to effectively filter LLM descriptions for better alignment. \textbf{ABS} achieves state-of-the-art performance on out-of-distribution generalization and zero-shot classification tasks. Notably, \textbf{ABS} is training-free and even rivals few-shot and test-time adaptation methods. Our code is available at \href{https://github.com/BIT-DA/ABS}{\textcolor{darkgreen}{https://github.com/BIT-DA/ABS}}.
Learning robust representations from data often requires scale, which has led to the success of recent zero-shot models such as CLIP. However, the obtained robustness can easily be deteriorated when these models are fine-tuned on other downstream tasks (e.g., of smaller scales). Previous works often interpret this phenomenon in the context of domain shift, developing fine-tuning methods that aim to preserve the original domain as much as possible. However, in a different context, fine-tuned models with limited data are also prone to learning features that are spurious to humans, such as background or texture. In this paper, we propose StarFT (Spurious Textual Alignment Regularization), a novel framework for fine-tuning zero-shot models to enhance robustness by preventing them from learning spuriosity. We introduce a regularization that aligns the output distribution for spuriosity-injected labels with the original zero-shot model, ensuring that the model is not induced to extract irrelevant features further from these descriptions. We leverage recent language models to get such spuriosity-injected labels by generating alternative textual descriptions that highlight potentially confounding features. Extensive experiments validate the robust generalization of StarFT and its emerging properties: zero-shot group robustness and improved zero-shot classification. Notably, StarFT boosts both worst-group and average accuracy by 14.30% and 3.02%, respectively, in the Waterbirds group shift scenario, where other robust fine-tuning baselines show even degraded performance.
Pre-trained vision-language models such as contrastive language-image pre-training (CLIP) have demonstrated a remarkable generalizability, which has enabled a wide range of applications represented by zero-shot classification. However, vision-language models still suffer when they face datasets with large gaps from training ones, i.e., distribution shifts. We found that CLIP is especially vulnerable to sensor degradation, a type of realistic distribution shift caused by sensor conditions such as weather, light, or noise. Collecting a new dataset from a test distribution for fine-tuning highly costs since sensor degradation occurs unexpectedly and has a range of variety. Thus, we investigate test-time adaptation (TTA) of zero-shot classification, which enables on-the-fly adaptation to the test distribution with unlabeled test data. Existing TTA methods for CLIP mainly focus on modifying image and text embeddings or predictions to address distribution shifts. Although these methods can adapt to domain shifts, such as fine-grained labels spaces or different renditions in input images, they fail to adapt to distribution shifts caused by sensor degradation. We found that this is because image embeddings are "corrupted" in terms of uniformity, a measure related to the amount of information. To make models robust to sensor degradation, we propose a novel method called uniformity-aware information-balanced TTA (UnInfo). To address the corruption of image embeddings, we introduce uniformity-aware confidence maximization, information-aware loss balancing, and knowledge distillation from the exponential moving average (EMA) teacher. Through experiments, we demonstrate that our UnInfo improves accuracy under sensor degradation by retaining information in terms of uniformity.
Recent advances in vision language models (VLMs) have enabled broad progress in the general medical field. However, pathology still remains a more challenging subdomain, with current pathology specific VLMs exhibiting limitations in both diagnostic accuracy and reasoning plausibility. Such shortcomings are largely attributable to the nature of current pathology datasets, which are primarily composed of image description pairs that lack the depth and structured diagnostic paradigms employed by real world pathologists. In this study, we leverage pathology textbooks and real world pathology experts to construct high-quality, reasoning-oriented datasets. Building on this, we introduce Patho-R1, a multimodal RL-based pathology Reasoner, trained through a three-stage pipeline: (1) continued pretraining on 3.5 million image-text pairs for knowledge infusion; (2) supervised fine-tuning on 500k high-quality Chain-of-Thought samples for reasoning incentivizing; (3) reinforcement learning using Group Relative Policy Optimization and Decoupled Clip and Dynamic sAmpling Policy Optimization strategies for multimodal reasoning quality refinement. To further assess the alignment quality of our dataset, we propose PathoCLIP, trained on the same figure-caption corpus used for continued pretraining. Comprehensive experimental results demonstrate that both PathoCLIP and Patho-R1 achieve robust performance across a wide range of pathology-related tasks, including zero-shot classification, cross-modal retrieval, Visual Question Answering, and Multiple Choice Question. Our project is available at the Patho-R1 repository: https://github.com/Wenchuan-Zhang/Patho-R1.
Understanding the factors contributing to traffic crashes and developing strategies to mitigate their severity is essential. Traditional statistical methods and machine learning models often struggle to capture the complex interactions between various factors and the unique characteristics of each crash. This research leverages large language model (LLM) to analyze freeway crash data and provide crash causation analysis accordingly. By compiling 226 traffic safety studies related to freeway crashes, a training dataset encompassing environmental, driver, traffic, and geometric design factors was created. The Llama3 8B model was fine-tuned using QLoRA to enhance its understanding of freeway crashes and their contributing factors, as covered in these studies. The fine-tuned Llama3 8B model was then used to identify crash causation without pre-labeled data through zero-shot classification, providing comprehensive explanations to ensure that the identified causes were reasonable and aligned with existing research. Results demonstrate that LLMs effectively identify primary crash causes such as alcohol-impaired driving, speeding, aggressive driving, and driver inattention. Incorporating event data, such as road maintenance, offers more profound insights. The model's practical applicability and potential to improve traffic safety measures were validated by a high level of agreement among researchers in the field of traffic safety, as reflected in questionnaire results with 88.89%. This research highlights the complex nature of traffic crashes and how LLMs can be used for comprehensive analysis of crash causation and other contributing factors. Moreover, it provides valuable insights and potential countermeasures to aid planners and policymakers in developing more effective and efficient traffic safety practices.
Learning to associate audio with textual descriptions is valuable for a range of tasks, including pretraining, zero-shot classification, audio retrieval, audio captioning, and text-conditioned audio generation. Existing contrastive language-audio pretrained models are typically trained using global, clip-level descriptions, which provide only weak temporal supervision. We hypothesize that CLAP-like language-audio models - particularly, if they are expected to produce frame-level embeddings - can benefit from a stronger temporal supervision. To confirm our hypothesis, we curate a novel dataset of approximately 12,000 audio recordings from Freesound, each annotated with single-sentence free-text descriptions linked to a specific temporal segment in an audio recording. We use large language models to clean these annotations by removing references to non-audible events, transcribed speech, typos, and annotator language bias. We further propose a frame-wise contrastive training strategy that learns to align text descriptions with temporal regions in an audio recording and demonstrate that our model has better temporal text-audio alignment abilities compared to models trained only on global captions when evaluated on the AudioSet Strong benchmark. The dataset and our source code are available on Zenodo and GitHub, respectively.
In this paper, we propose a diffusion model that integrates a representation-conditioning mechanism, where the representations derived from a Vision Transformer (ViT) are used to condition the internal process of a Transformer-based diffusion model. This approach enables representation-conditioned data generation, addressing the challenge of requiring large-scale labeled datasets by leveraging self-supervised learning on unlabeled data. We evaluate our method through a zero-shot classification task for hematoma detection in brain imaging. Compared to the strong contrastive learning baseline, DINOv2, our method achieves a notable improvement of +6.15% in accuracy and +13.60% in F1-score, demonstrating its effectiveness in image classification.
Medical vision-language models (VLMs) have shown promise as clinical assistants across various medical fields. However, specialized dermatology VLM capable of delivering professional and detailed diagnostic analysis remains underdeveloped, primarily due to less specialized text descriptions in current dermatology multimodal datasets. To address this issue, we propose MM-Skin, the first large-scale multimodal dermatology dataset that encompasses 3 imaging modalities, including clinical, dermoscopic, and pathological and nearly 10k high-quality image-text pairs collected from professional textbooks. In addition, we generate over 27k diverse, instruction-following vision question answering (VQA) samples (9 times the size of current largest dermatology VQA dataset). Leveraging public datasets and MM-Skin, we developed SkinVL, a dermatology-specific VLM designed for precise and nuanced skin disease interpretation. Comprehensive benchmark evaluations of SkinVL on VQA, supervised fine-tuning (SFT) and zero-shot classification tasks across 8 datasets, reveal its exceptional performance for skin diseases in comparison to both general and medical VLM models. The introduction of MM-Skin and SkinVL offers a meaningful contribution to advancing the development of clinical dermatology VLM assistants. MM-Skin is available at https://github.com/ZwQ803/MM-Skin
Contrastive Language-Image Pre-training (CLIP) excels in multimodal tasks such as image-text retrieval and zero-shot classification but struggles with fine-grained understanding due to its focus on coarse-grained short captions. To address this, we propose Fine-Grained CLIP (FG-CLIP), which enhances fine-grained understanding through three key innovations. First, we leverage large multimodal models to generate 1.6 billion long caption-image pairs for capturing global-level semantic details. Second, a high-quality dataset is constructed with 12 million images and 40 million region-specific bounding boxes aligned with detailed captions to ensure precise, context-rich representations. Third, 10 million hard fine-grained negative samples are incorporated to improve the model's ability to distinguish subtle semantic differences. We construct a comprehensive dataset, termed FineHARD, by integrating high-quality region-specific annotations with hard fine-grained negative samples. Corresponding training methods are meticulously designed for these data. Extensive experiments demonstrate that FG-CLIP outperforms the original CLIP and other state-of-the-art methods across various downstream tasks, including fine-grained understanding, open-vocabulary object detection, image-text retrieval, and general multimodal benchmarks. These results highlight FG-CLIP's effectiveness in capturing fine-grained image details and improving overall model performance. The data, code, and models are available at https://github.com/360CVGroup/FG-CLIP.
In this work, we seek to understand the performance of large language models in the mechanical engineering domain. We leverage the semantic data found in the ABC dataset, specifically the assembly names that designers assigned to the overall assemblies, and the individual semantic part names that were assigned to each part. After pre-processing the data we developed two unsupervised tasks to evaluate how different model architectures perform on domain-specific data: a binary sentence-pair classification task and a zero-shot classification task. We achieved a 0.62 accuracy for the binary sentence-pair classification task with a fine-tuned model that focuses on fighting over-fitting: 1) modifying learning rates, 2) dropout values, 3) Sequence Length, and 4) adding a multi-head attention layer. Our model on the zero-shot classification task outperforms the baselines by a wide margin, and achieves a top-1 classification accuracy of 0.386. The results shed some light on the specific failure modes that arise when learning from language in this domain.
Vision-language models (VLMs) allow to embed texts and images in a shared representation space. However, it has been shown that these models are subject to a modality gap phenomenon meaning there exists a clear separation between the embeddings from one modality and another in the embedding space. While this misalignment is detrimental for downstream tasks such as multimodal retrieval, multimodal clustering or zero-shot classification, etc. no generic and practical methods have so far been proposed to assess it precisely and even reduce it. We therefore propose novel measures and effective techniques (spectral- and optimal transport-based methods) to achieve this goal. Extensive experiments conducted on several image-text datasets and models demonstrate their effectiveness and beneficial effects on downstream tasks. Our code is available at the URL provided in the paper's abstract.
This paper investigates hybrid intelligence and collaboration between researchers of sexism and Large Language Models (LLMs), with a four-component pipeline. First, nine sexism researchers answer questions about their knowledge of sexism and of LLMs. They then participate in two interactive experiments involving an LLM (GPT3.5). The first experiment has experts assessing the model's knowledge about sexism and suitability for use in research. The second experiment tasks them with creating three different definitions of sexism: an expert-written definition, an LLM-written one, and a co-created definition. Lastly, zero-shot classification experiments use the three definitions from each expert in a prompt template for sexism detection, evaluating GPT4o on 2.500 texts sampled from five sexism benchmarks. We then analyze the resulting 67.500 classification decisions. The LLM interactions lead to longer and more complex definitions of sexism. Expert-written definitions on average perform poorly compared to LLM-generated definitions. However, some experts do improve classification performance with their co-created definitions of sexism, also experts who are inexperienced in using LLMs.