arXiv subjects

Elisa Ricci

Publications and source records attributed to Elisa Ricci.

At least 19 recordsLinked to original sources

arXiv2026

Hierarchical Step Decomposition for Training-free Online Video Step Grounding

Given a set of procedural steps, Video Step Grounding (VSG) aims to identify which steps occur in a video and localize when they are performed. Existing methods are either offline, requiring access to the full video before making predictions, or online, but relying on fixed-length video segments producing noisy, over-segmented outputs. In this work, we propose Hierarchical Step Decomposition (HSD), a training-free framework for online VSG that leverages the semantic understanding capabilities of Large Multimodal Models. HSD addresses two central challenges in online grounding: constructing semantically coherent temporal segments from a streaming video, and bridging the semantic gap between coarse procedural step descriptions and visual observations. To construct elementary temporal units, HSD incrementally groups incoming segments into semantically consistent events, using an anchor caption that is continuously validated against the visual stream. Based on these events, we perform zero-shot procedural step classification and selectively refine uncertain predictions by decomposing candidate steps into finegrained substeps that capture more nuanced visual evidence. By operating on semantically coherent events rather than every fixed-length segment, HSD reduces redundant computation and produces more stable predictions. Experiments on HT-Step, Ego4D Goal-Step, and CrossTask show that HSD substantially outperforms state-of-the-art methods, overall achieving better step recognition, more accurate temporal localization, and faster inference.

cs.CV↗

arXiv2026

Video-Language Entailment for Video Similarity Estimation

Comparing videos to assess their semantic similarity remains a challenging task for current Large Multimodal Models (LMMs), which are primarily trained with single-video objectives. While most prior works have focused on computing global similarity, recent studies have introduced the more challenging problem of concept-based similarity, which measures how similar two videos are with respect to specific concepts (e.g., activity, location). However, no dedicated approaches have yet leveraged LMMs to explicitly address this task. In this paper, we aim to fill this gap introducing Video Caption Entailment for video similarity estimation (VICE): a dual-branch design that leverages LMMs for cross-video reasoning. Each branch jointly processes one video and the caption of the other, measuring their alignment for a specific concept via video-language entailment. The outputs from both branches are fused to produce a final similarity score, along with a textual explanation of the model reasoning. This approach leverages the model's native single-video understanding while enabling fine-grained comparative reasoning between video and text. Experiments on ConViS-Bench and StepDiff show that VICE outperforms models that process both videos simultaneously, highlighting the promise of single-video reasoning for concept-based similarity estimation.

cs.CV↗

arXiv2026

Localize Any Object in X-Ray Security Scans without Human Annotation

Universal object localization in X-ray security inspection is critical for automated threat detection in safety-critical venues. However, unlike everyday RGB images that dominate web-scale visual data, X-ray scans exhibit distinct color patterns, ambiguous boundaries, and compositional structures caused by volumetric superposition. These gaps hinder the direct zero-shot transfer of dense perception foundation models trained on web-scale RGB data. Moreover, annotated X-ray data is scarce and requires expert labeling, limiting both the training of generalizable X-ray native models and the adaptation of RGB foundation models for X-ray data via fine-tuning. Given these challenges, the bright promise of highly generalizable perception models, enabled by data scaling laws in the RGB domain, remains largely out of reach for X-ray inspection. To this end, we introduce LAO-X, a self-supervised adaptation framework that Locates Any Object in X-ray scans using diverse synthesized image--annotation pairs with granularity-aware supervision. LAO-X first designs a saliency-guided X-ray object mining module to separate diverse object instances, which are then used for physics-guided synthesis in the absorbance domain. LAO-X further incorporates an occlusion-controlled curriculum strategy to fine-tune a Segment Anything Model 2 (SAM2) localizer, progressively adapting it to X-ray scans with increasing object counts and overlap levels. Experiments on six X-ray benchmarks show that LAO-X substantially improves category-agnostic localization, achieving 2\% to 23\% mAP gains over SAM2 and X-ray specific baselines in heavily cluttered scenarios, entirely without human-annotated labels.

cs.CV↗

arXiv2026

Kinematics-Centric Continuous Sign Language Retrieval with Gloss-Guided Boundary-Aware Alignment

Sign language-text alignment remains a fundamental challenge for text-driven sign language understanding. Existing methods predominantly rely on appearance-heavy RGB representations, which entangle motion semantics with visual variations and lead to ambiguous motion-language grounding. In this paper, we reformulate sign language-text alignment in a structured kinematic space and propose a kinematics-centric framework that adopts 3D SMPL-X motion as the primary representation. By explicitly modeling the kinematic dynamics of signing in a unified motion space, our approach reduces reliance on appearance signals and yields more semantically consistent representations. To capture the compositional nature of sign language, we introduce a gloss-guided local alignment mechanism that leverages gloss temporal spans as weak supervision to decompose continuous motion into coherent segments and establish fine-grained motion-text correspondences, thereby reducing ambiguity in localizing word-level semantics in continuous signing. Furthermore, we develop a visual distillation strategy, where RGB signals serve as privileged supervision during training to provide complementary contextual cues, while being completely removed at inference time. Extensive experiments on standard benchmarks demonstrate that our method achieves state-of-the-art bidirectional retrieval performance on CSL-Daily and competitive results on PHOENIX-2014T. These results highlight the effectiveness of kinematic representations and explicit local grounding for sign language-text alignment.

cs.CV↗

arXiv2026

Spot-the-shift: Evaluating Grounded Image Difference Captioning of Long-term Changes

Long-term change understanding from images of the same place revisited over time is a challenging task with applications in map maintenance and urban infrastructure monitoring. Prior work addresses it either through pixel-level prediction or difference captioning, neither of which is sufficient to reliably measure how well models detect and describe such changes. We introduce SPOT-THE-SHIFT, a human-verified benchmark for grounded image difference captioning of long-term changes in real-world driving scenes. Our benchmark provides natural language captions and spatial masks for structural changes across each image pair. We further propose an evaluation protocol that reliably assesses models' captioning ability, validated through human studies. Benchmarking state-of-the-art MLLMs, we find that models struggle with the fine-grained multi-image spatial capability required for this task. Finally, we develop a synthetic data generation pipeline that improves an off-the-shelf MLLM without sacrificing general capabilities.

cs.CV↗

arXiv2026

How (Mis)calibrated is your Federated CLIP and what to do about it?

Vision-language models (VLMs) such as CLIP are increasingly adapted across decentralized data silos, yet the reliability of their predictions under federated learning (FL) remains largely unexplored. In this work, we present a systematic study of calibration in federated CLIP under non-IID client distributions. Our experiments reveal that widely used prompt-tuning methods consistently degrade calibration, often yielding substantially higher calibration error despite competitive recognition performance, while explicit training-time calibration regularizers provide only limited improvements. Motivated by these findings, we identify the choice of fine-tuning parameterization as a critical factor governing calibration and conduct a controlled comparison between prompt tuning and five backbone fine-tuning (BFT) strategies: AdaptFormer, LayerNorm, LoRA, VeRA, and DoRA. We find that BFT methods generally offer a more favorable accuracy-calibration trade-off than prompt tuning, although their benefits are not universal. Through extensive analysis, we show that calibration behavior is closely linked to the geometry of federated updates, residual parameterization, and the resulting client and logit drift. Across in-distribution, domain-generalization, and base-to-new evaluation settings, our results establish fine-tuning parameterization as a central design choice for building accurate and reliable federated CLIP models. Codes are available at https://github.com/mainaksingha01/FL2oRA.

cs.CV↗

arXiv2026

FineHOI: Part-Aware Dense Representations for Zero-Shot Human-Object Interaction Detection

Human-Object Interaction (HOI) detection aims to localize humans and objects in images and classify their interactions. Zero-shot HOI focuses on recognizing interactions that are not observed during training, requiring models to generalize beyond seen verb-object compositions. Recent approaches leverage Vision-Language Models (VLMs), benefiting from rich semantic representations. However, they often rely on global or detector-centric features that compress interaction cues and hinder fine-grained spatial reasoning. To overcome this limitation, we propose FineHOI, a zero-shot HOI framework that explicitly models interactions from dense patch-level features. Our approach is motivated by the observation that human-object interactions are defined by localized spatial relationships, which are not preserved by global and detector-centric representations. To this end, we introduce an Adaptive Part-Level Attention module that decomposes humans and objects into semantically coherent parts via unsupervised clustering, and re-weights them based on their interaction relevance. These representations are then integrated through a Region-Aware Interaction Transformer that integrates part-aware and global features and produces the final HOI embedding. Extensive experiments demonstrate that FineHOI consistently outperforms existing zero-shot HOI methods, achieving particularly strong gains on unseen interactions. Code is available at https://github.com/francescotonini/fine-hoi.

cs.CV↗

arXiv2026

DataComp-VLM: Improved Open Datasets for Vision-Language Models

Building performant Vision-Language Models (VLMs) requires carefully curating large-scale training datasets, yet the community lacks systematic benchmarks for evaluating such curation strategies. We introduce DataComp for VLMs (DCVLM), a benchmark for controlled data-centric experiments to improve VLM training. As part of DCVLM, we collect 160 datasets spanning four data types -- image-caption pairs, multimodal interleaved documents, text-only, and instruction-tuning data -- into a corpus of 6T multimodal tokens. DCVLM allows participants to test curation strategies (filtering, mixing, formatting, sampling) across 1B-8B models and 6.25B-200B token budgets. Models are then evaluated on a carefully selected suite of up to 52 downstream benchmarks across 9 domains. We conduct extensive experiments on DCVLM and find that data mixing, not filtering, is key to a high-quality training dataset: instruction-heavy mixtures scale better than caption-heavy ones, with gains widening at larger scales. The resulting dataset, DCVLM-Baseline, enables training an 8B VLM to 63.6% accuracy on our 33-task core suite with 200B training tokens. Compared to FineVision, the state-of-the-art open VLM training dataset, this represents an improvement of +5.4pp. DCVLM and all accompanying artifacts will be made publicly available at https://www.datacomp.ai/dcvlm/.

cs.CV↗

arXiv2026

DistMoE: Private-data Rehearsal-free Routing in Mixture-of-Experts for Distributed Instruction Tuning

Multimodal Large Language Models (MLLMs) have shown strong multimodal instruction-following ability, but adapting them to diverse visual-language domains typically assumes centralized data access and costly joint training. This is restrictive when data is distributed across private, domain-specific, or permission-limited clients. To this end, we propose DistMoE, a mixture-of-experts (MoE) approach for distributed visual instruction tuning. In each layer of the language decoder it augments the public feedforward network (FFN) with a client-specific private FFN expert, with the goal to acquire domain-specific knowledge. However, independent expert training causes the private FFNs to learn representation of different scale and magnitudes, making merging the experts difficult. To reduce client-specific drift, we introduce a public-anchored expert composition stage that updates only routers and lightweight private projection adapters on a mix of local client data and public data, via an isotropic regularization loss, therefore making it cross-client rehearsal-free composition. During inference, DistMoE performs modular routing over public and private experts, enabling token-wise domain composition without explicit domain labels. Experiments across diverse visual-language benchmarks show that DistMoE enables flexible expert reuse, effective domain adaptation, and competitive performance while preserving modular control over client-specific knowledge. Codes are available at https://github.com/mainaksingha01/DistMoE.

cs.CV↗

arXiv2026

Unlearning Under Imbalance: Benchmarking Fairness in Multimodal LLM Unlearning

Machine unlearning has emerged as a tool for removing personal data from trained models to comply with recent AI regulations. To evaluate unlearning effectiveness in multimodal large language models (MLLMs), prior works fine-tune models on fictitious identities, simulating unlearning requests on subsets of these IDs, which are typically uniformly distributed. However, in realistic scenarios, people from different demographic groups may request to be unlearned at different frequencies, potentially altering the model's internal beliefs for these groups and leading to biased behaviors. To fill this gap, we propose FAIRGET, the first Visual Question Answering benchmark that evaluates unlearning under unbalanced, realistic, forget requests. These requests are designed to simulate multiple realistic scenarios, ranging from simple to challenging settings, that lead to biased unlearned models if fairness is not accounted for. Additionally, we propose FAUN, the first unlearning algorithm for MLLMs that forgets unlearning data while preserving model fairness. FAUN exploits a bias-aware activation steering mechanism to unlearn identities while accounting for the unbalanced nature of the forget data. Experiments on FAIRGET and the established FIUBench demonstrate our method's superiority both in unlearning quality and fairness.

cs.CV↗

arXiv2026

Personalizing MLLMs via Reinforced Multimodal Reference Game

Personalizing Multimodal Large Language Models (MLLMs) aims to recognize users' unique concepts from visual data and provide personalized responses. Although prior work has shown the benefit of concept descriptions and reasoning for this task, MLLM descriptions often include information, such as state and context, that does not help and may in fact hinder the unique identification of the target concept among other visually similar items. Effective descriptions of personal concepts should instead be accurate, discriminative, and free of distracting details. To achieve such descriptions, we introduce Reinforced Reference Game (RRG), a learning framework that promotes discriminative descriptions through a novel reinforced multimodal reference game. The MLLM plays both the roles of speaker and listener in a contrastive game setting, whose goal is to effectively communicate discriminative information about a target concept. Our approach formulates a verifiable contrastive reward over hard positives (dissimilar views of the same concept) and hard negatives (visually similar but different concepts). Empirically, RRG achieves state-of-the-art across multiple tasks on three personalization benchmarks. RRG generalizes to unseen domains and outperforms existing methods based on concept descriptions and personalization-specific RL frameworks. We will release code and models in the project page.

cs.CV↗

arXiv2026

ProactiveBench: Benchmarking Proactiveness in Multimodal Large Language Models

Effective collaboration begins with knowing when to ask for help. For example, when trying to identify an occluded object, a human would ask someone to remove the obstruction. Can MLLMs exhibit a similar "proactive" behavior by requesting simple user interventions? To investigate this, we introduce ProactiveBench, a benchmark built from seven repurposed datasets that tests proactiveness across different tasks such as recognizing occluded objects, enhancing image quality, and interpreting coarse sketches. We evaluate 22 MLLMs on ProactiveBench, showing that (i) they generally lack proactiveness; (ii) proactiveness does not correlate with model capacity; (iii) "hinting" at proactiveness yields only marginal gains. Surprisingly, we found that conversation histories and in-context learning introduce negative biases, hindering performance. Finally, we explore a simple fine-tuning strategy based on reinforcement learning: its results suggest that proactiveness can be learned, even generalizing to unseen scenarios. We publicly release ProactiveBench as a first step toward building proactive multimodal models.

cs.CV↗

arXiv2026

Position: Modular Memory is the Key to Continual Learning Agents

Foundation models have transformed machine learning through large-scale pretraining and increased test-time compute. Despite surpassing human performance in several domains, these models remain fundamentally limited in continuous operation, experience accumulation, and personalization, capabilities that are central to adaptive intelligence. While continual learning research has long targeted these goals, its historical focus on in-weight learning (IWL), i.e., updating a single model's parameters to absorb new knowledge, has rendered catastrophic forgetting a persistent challenge. Our position is that combining the strengths of In-Weight Learning (IWL) and the newly emerged capabilities of In-Context Learning (ICL) through the design of modular memory is the missing piece for continual adaptation at scale. We outline a conceptual framework for modular memory-centric architectures that leverage ICL for rapid adaptation and knowledge accumulation, and IWL for stable updates to model capabilities, charting a practical roadmap toward continually learning agents.

cs.LG↗

arXiv2026

Thinking Past the Answer: Evaluating Harmful Overthinking in Large Reasoning Models

Large Reasoning Models (LRMs) improve performance by generating explicit intermediate reasoning traces through increased test-time compute, yet the assumption that longer reasoning is consistently beneficial remains under-examined. While recent evidence shows that additional reasoning can lead models to overthink, we ask: "Once a model has reached the correct answer, does further reasoning refine the solution, or deviate from it?" To study the dynamics after correctness, we introduce a prefix-level trajectory evaluation protocol grounded in reasoning sufficiency, defining the minimum reasoning budget required for a model to first generate the correct answer. This allows us to disentangle verbose overthinking, where additional reasoning is redundant but harmless, from harmful overthinking, where continued reasoning destabilizes an already-correct trajectory. Starting from multimodal benchmarks, we find that many instances considered reasoning-intensive require surprisingly little reasoning. Moreover, stopping at the first correct prefix improves accuracy over standard reasoning up to 21%, revealing that current models are limited not only by their ability to reason, but also by their inability to stop at the right time. Furthermore, while common efficiency strategies like early stopping substantially reduce verbose overthinking (up to 50%), they fail to mitigate harmful overthinking. Failure analysis reveals that correctness deviations are mainly driven by logical drift and visual reinterpretation. Finally, we show that our findings generalize to language-only reasoning benchmarks, highlighting harmful overthinking as a broader reliability risk. Code available at https://simonecaldarella.github.io/thinking-past-the-answer.

cs.AI↗

arXiv2026

Zero-Shot Temporal Action Localization Through Textual Guidance

Zero-shot temporal action localization (ZS-TAL) consists of classifying and localizing actions in untrimmed videos, where action classes are unseen at training time. Existing work uses Vision and Language Models (VLMs), taking advantage of their strong zero-shot transfer capabilities. Yet, these models face evident challenges with fine-grained action classification, making it difficult to directly use them to distinguish between the presence and absence of an action. Most current methods for ZS-TAL address these challenges by training models on large-scale video datasets, which require annotated data and often result in limited generalization performance. Recently, approaches discarding the use of labeled data have emerged as an alternative. Following this direction, we propose a novel approach, ``Textual Guidance for finer localization of actions in videos'' (TEGU), that compensates for the lack of supervision from training data by exploiting rich textual information derived from large language models and structured text extracted from captions. This additional linguistic context can improve fine-grained discrimination by providing richer cues about fine-grained action differences within videos. We validate the effectiveness of the proposed method by conducting experiments on the THUMOS14 and the ActivityNet-v1.3 datasets. Our results show that, by exploiting rich textual information for improved action localization, TEGU outperforms state-of-the-art ZS-TAL approaches that do not involve training

cs.CV↗

arXiv2026

Text-to-CAD Evaluation with CADTests

Text-to-CAD has recently emerged as an important task with the potential to substantially accelerate design workflows. Despite its significance, there has been surprisingly little work on Text-to-CAD evaluation, and assessing CAD model generation performance remains a considerable challenge. In this work, we introduce a new evaluation perspective for Text-to-CAD based on automated testing. We propose CADTestBench, the first test-based benchmark for Text-to-CAD, based on CADTests, executable software tests that verify whether a generated CAD model satisfies the geometric and topological requirements of the input prompt. Using CADTestBench, we conduct comprehensive benchmarking of recent Text-to-CAD methods and further demonstrate that CADTests can also guide CAD model generation, yielding simple baselines that surpass performance of current methods. CADTestBench code and data are available at GitHub and Hugging Face dataset.

cs.CV↗

arXiv2026

Ensembling Pruned Attention Heads For Uncertainty-Aware Efficient Transformers

Uncertainty quantification (UQ) is essential for deploying deep neural networks in safety-critical settings. Although methods like Deep Ensembles achieve strong UQ performance, their high computational and memory costs hinder scalability to large models. We introduce Hydra Ensembles, an efficient transformer-based ensemble that prunes attention heads to create diverse members and merges them via a new multi-head attention with grouped fully-connected layers. This yields a compact model with inference speed close to a single network, matching or surpassing Deep Ensembles in UQ performance without retraining from scratch. We also provide an in-depth analysis of pruning, showing that naive approaches can harm calibration, whereas Hydra Ensembles preserves robust uncertainty. Experiments on image and text classification tasks, with various architectures, show consistent gains over Deep Ensembles. Remarkably, in zero-shot classification on ImageNet-1k, our approach surpasses state of the art methods, even without requiring additional training.

cs.LG↗

arXiv2026

CLIPoint3D: Language-Grounded Few-Shot Unsupervised 3D Point Cloud Domain Adaptation

Recent vision-language models (VLMs) such as CLIP demonstrate impressive cross-modal reasoning, extending beyond images to 3D perception. Yet, these models remain fragile under domain shifts, especially when adapting from synthetic to real-world point clouds. Conventional 3D domain adaptation approaches rely on heavy trainable encoders, yielding strong accuracy but at the cost of efficiency. We introduce CLIPoint3D, the first framework for few-shot unsupervised 3D point cloud domain adaptation built upon CLIP. Our approach projects 3D samples into multiple depth maps and exploits the frozen CLIP backbone, refined through a knowledge-driven prompt tuning scheme that integrates high-level language priors with geometric cues from a lightweight 3D encoder. To adapt task-specific features effectively, we apply parameter-efficient fine-tuning to CLIP's encoders and design an entropy-guided view sampling strategy for selecting confident projections. Furthermore, an optimal transport-based alignment loss and an uncertainty-aware prototype alignment loss collaboratively bridge source-target distribution gaps while maintaining class separability. Extensive experiments on PointDA-10 and GraspNetPC-10 benchmarks show that CLIPoint3D achieves consistent 3-16% accuracy gains over both CLIP-based and conventional encoder-based baselines. Project page: https://sarthakm320.github.io/CLIPoint3D.

cs.CV↗