Search arXiv⌕ Search

arXiv subjects

Min Cao

Publications and source records attributed to Min Cao.

At least 19 recordsLinked to original sources
arXiv2026

Navigating Sparse Evidence: Agentic Visual RAG via Explicit Context Selection and Consolidation

Visual Retrieval-Augmented Generation (VRAG) empowers models to navigate and answer queries about visually rich documents by retrieving relevant page images as visual evidence and reasoning over their content. However, effectively utilizing this visual evidence is usually impeded by two main challenges. First, answer-relevant evidence is sparse and may be concentrated in a small region of one page or dispersed across multiple pages. Second, existing agentic methods often generate answers based on raw exploration trajectories or compressed textual memories rather than an explicitly organized set of supporting images, making answers susceptible to exploration noise and obscuring the evidence-backed reasoning trace. We argue that the bottleneck lies not only in evidence discovery but also in its preservation and organization before answer generation. We propose SCoRE (Selection and Consolidation for Robust Evidence), a unified agent loop for explicit evidence selection and consolidation. During exploration, SCoRE retains only query-relevant observations and their source pointers in a maintained textual ledger, preserving earlier evidence while keeping the visual context bounded. At termination, it reloads the referenced original images and consolidates the visual evidence for answering, arranging it into a logical sequence. This decouples final reasoning from exploratory trial-and-error while ensuring strict visual grounding via indexed claim-to-image linkages. To enable end-to-end optimization of this unified rollout, our training paradigm combines filtered cold-start trajectory distillation with evidence-aware reinforcement learning, whose reward promotes evidence coverage, consolidation compactness, and answer correctness.

cs.AI↗
arXiv2026

Generative Retrieval for Unsupervised Text-Based Person Search

Text-based person search (TBPS) aims to retrieve images of a target person from a large image gallery based on a given natural language description. Most existing methods rely on supervised learning with manually annotated image-text pairs. In this paper, we explore unsupervised TBPS, with only unlabeled images. We propose GTR+, a two-stage generation-then-retrieval framework. In the generation stage, we introduce a tiered description generation framework designed to produce fine-grained and stylistically diverse textual descriptions through a three-tier sequential process. The base tier leverages an automated question-and-answer mechanism to generate basic visual attribute descriptions; the intermediate tier enhances fine-grained detail using an inter-sample contrastive mechanism; the advanced tier further enriches textual diversity via a stylized expansion mechanism. In the retrieval stage, to mitigate the impact of noisy pseudo texts, we develop an adaptive confidence-weighted retrieval learning framework. We model image-text pairs as clean or noisy using a Gaussian Mixture Model, calibrated by real-time image-text similarity and static text generation probability from the prior stage, yielding adaptive sample weights during training. Beyond that, we also contribute LargeFine-Person, a large-scale TBPS dataset with high-quality, fine-grained, and diverse textual annotations, enabling a practical and generalizable TBPS pre-training benchmark under unsupervised setting. Experiments on multiple TBPS benchmarks demonstrate the effectiveness and generalization of both GTR+ and LargeFine-Person. Code is available at: https://github.com/Flame-Chasers/GTR.

cs.CV↗
arXiv2026

VISOR: Agentic Visual Retrieval-Augmented Generation via Iterative Search and Over-horizon Reasoning

Visual Retrieval-Augmented Generation (VRAG) empowers Vision-Language Models to retrieve and reason over visually rich documents. To tackle complex queries requiring multi-step reasoning, agentic VRAG systems interleave reasoning with iterative retrieval. However, existing agentic VRAG faces two critical bottlenecks. (1) Visual Evidence Sparsity: key evidence is scattered across pages yet processed in isolation, hindering cross-page reasoning; moreover, fine-grained intra-image evidence often requires precise visual actions, whose misuse degrades retrieval quality; (2) Search Drift in Long Horizons: the accumulation of visual tokens across retrieved pages dilutes context and causes cognitive overload, leading agents to deviate from their search objective. To address these challenges, we propose VISOR (Visual Retrieval-Augmented Generation via Iterative Search and Over-horizon Reasoning), a unified single-agent framework. VISOR features a structured Evidence Space for progressive cross-page reasoning, coupled with a Visual Action Evaluation and Correction mechanism to manage visual actions. Additionally, we introduce a Dynamic Trajectory with Sliding Window and Intent Injection to mitigate search drift. They anchor the evidence space while discarding earlier raw interactions, preventing context from being overwhelmed by visual tokens. We train VISOR using a Group Relative Policy Optimization-based Reinforcement Learning (GRPO-based RL) pipeline with state masking and credit assignment tailored for dynamic context reconstruction. Extensive experiments on ViDoSeek, SlideVQA, and MMLongBench demonstrate that VISOR achieves state-of-the-art performance while maintaining reasonable and controllable computational costs for long-horizon visual reasoning tasks. The source code is available at https://github.com/syc1336/VISOR.

cs.CV↗
arXiv2026

Facial-R1: Aligning Reasoning and Recognition for Facial Emotion Analysis

Facial Emotion Analysis (FEA) extends traditional facial emotion recognition by incorporating explainable, fine-grained reasoning. The task integrates three subtasks: emotion recognition, facial Action Unit (AU) recognition, and AU-based emotion reasoning to model affective states jointly. While recent approaches leverage Vision-Language Models (VLMs) and achieve promising results, they face two critical limitations: (1) hallucinated reasoning, where VLMs generate plausible but inaccurate explanations due to insufficient emotion-specific knowledge; and (2) misalignment between emotion reasoning and recognition, caused by fragmented connections between observed facial features and final labels. We propose Facial-R1, a three-stage alignment framework that effectively addresses both challenges with minimal supervision. First, we employ instruction fine-tuning to establish basic emotional reasoning capability. Second, we introduce reinforcement training guided by emotion and AU labels as reward signals, which explicitly aligns the generated reasoning process with the predicted emotion. Third, we design a data synthesis pipeline that iteratively leverages the prior stages to expand the training dataset, enabling scalable self-improvement of the model. Built upon this framework, we introduce FEA-20K, a benchmark dataset comprising 17,737 training and 1,688 test samples with fine-grained emotion analysis annotations. Extensive experiments across eight standard benchmarks demonstrate that Facial-R1 achieves state-of-the-art performance in FEA, with strong generalization and robust interpretability.

cs.CV↗
arXiv2026

Bounded-Compute Multimodal Regression for Product-Rating Prediction

Vision-language models (VLMs) are increasingly attractive for multimodal quality assessment, but their default reliance on autoregressive text generation and dynamic visual processing is poorly matched to scalar regression under strict latency budgets. We present a bounded-compute adaptation of SmolVLM2-256M-Video-Instruct for product-rating prediction in the LoViF 2026 Efficient VLM challenge. Motivated by recent multimodal engagement-prediction results showing that feature-based regression can outperform token-based score generation, we replace the language-modeling head with a lightweight two-layer MLP fed by pooled decoder states, and we enforce deterministic inputs through fixed 384x384 images and truncated metadata. Across controlled ablations, static global image processing slightly outperforms dynamic tiling, and scaling from 100K to 16M training examples substantially improves validation correlation. Under the official held-out evaluation, our 228M-parameter model achieves 0.39 PLCC and 0.40 CES, providing a strong and reproducible baseline for resource-constrained multimodal regression.

cs.CV↗
arXiv2026

ProCrit: Self-Elicited Multi-Perspective Reasoning with Critic-Guided Revision for Multimodal Sarcasm Detection

Multimodal sarcasm detection requires reasoning over cross-modal incongruities between literal expression and intended meaning, yet the specific analytical perspectives needed vary across samples due to the diversity of sarcastic mechanisms. While recent methods make this analytical process explicit, they still rely on fixed, predefined perspectives that operate independently under hand-crafted routing rules. We argue that multimodal sarcasm detection instead calls for self-elicited multi-perspective reasoning, where a model autonomously generates the perspectives needed for each sample and progressively integrates them into a coherent analysis. To realize this goal, we propose ProCrit, a Proposal-Critic two-agent framework with a proposal agent for multi-perspective reasoning and a critic agent for external evaluation and targeted revision guidance. First, to overcome the lack of process-level supervision in existing sarcasm datasets, ProCrit synthesizes process-level reasoning annotations through a dynamic-role agentic rollout: a strong vision-language model sequentially spawns analytical roles within a shared context, and the resulting multi-role trajectories are flattened into sequences that preserve cross-perspective dependencies while enabling efficient autoregressive generation. Second, to improve reasoning reliability, ProCrit adopts a draft-critique-revise paradigm in which an independent critic identifies reasoning deficiencies and provides targeted natural-language feedback for directed revision. Finally, we develop a mutual-refinement training framework that jointly optimizes proposal drafting and feedback-guided revision via dual-stage reinforcement learning, while refining the critic agent according to the actual effectiveness of its feedback. Experiments on three widely used benchmarks demonstrate the effectiveness of ProCrit.

cs.MA↗
arXiv2026

AsymTalker: Identity-Consistent Long-Term Talking Head Generation via Asymmetric Distillation

Diffusion-based talking head generation has achieved remarkable visual quality, yet scaling it to long-term videos remains challenging. The widely adopted chunk-wise paradigm introduces two fundamental failures: (1) temporal-spatial misalignment between static identity references and dynamic audio streams, and (2) cascading identity drift propagated through self-generated continuity references across chunks. To address both issues, we propose AsymTalker, a novel diffusion-based talking head generation method comprising Temporal Reference Encoding (TRE) and Asymmetric Knowledge Distillation (AKD). First, TRE mitigates temporal-spatial misalignment by transforming the static identity image into a temporally coherent latent representation through encoding of a temporally replicated pseudo-video, without introducing additional parameters. Second, AKD resolves the inherent conditioning dilemma in chunk-wise training: using ground-truth references causes train-inference mismatch, while self-generated references entangle supervision with identity drift. Our asymmetric design circumvents this by anchoring the teacher model with ground-truth continuity references to provide drift-free, chunk-level supervision, thereby avoiding the teacher bottleneck. Meanwhile, the student model learns under inference-aligned conditions, conditioned only on self-generated references, and is trained via distribution matching to preserve identity over long horizons. Extensive experiments show AsymTalker achieves state-of-the-art results on HDTF and VFHQ. It guarantees high-fidelity, identity-consistent synthesis over 600-second videos and reaches a real-time inference speed of 66 FPS.

cs.LG↗
arXiv2026

An Empirical Study of Validating Synthetic Data for Text-Based Person Retrieval

Data plays a pivotal role in Text-Based Person Retrieval (TBPR) research. Mainstream research paradigm necessitates real-world person images with manual textual annotations for training models, posing privacy concerns and annotation burdens. Several pioneering efforts explore synthetic data generation, and yet still depend on real data as a foundation, inheriting the same limitations. The feasibility of purely synthetic TBPR data remains unexplored, and there is currently no systematic study on the effectiveness boundaries of synthetic data across various real-world scenarios. In this work, we present the first comprehensive empirical study of synthetic data for TBPR, with two key aspects. (1) We propose a unified data synthesis pipeline that can operate entirely without real person data. It combines an inter-class image generation module that produces diverse identity-centric images by means of an automatic prompt construction strategy, and an intra-class augmentation module that enhances identity variation through text-driven image editing. (2) Leveraging this pipeline and an automatic textual description generation, we explore the effectiveness of synthetic data in diverse scenarios through extensive experiments, to reveal its practical utility as either a standalone replacement or a complementary augmentation to real data.

cs.CV↗
arXiv2026

GigaWorld-Policy: An Efficient Action-Centered World--Action Model

World-Action Models (WAM) initialized from pre-trained video generation backbones have demonstrated remarkable potential for robot policy learning. However, existing approaches face two critical bottlenecks that hinder performance and deployment. First, jointly reasoning over future visual dynamics and corresponding actions incurs substantial inference overhead. Second, joint modeling often entangles visual and motion representations, making motion prediction accuracy heavily dependent on the quality of future video forecasts. To address these issues, we introduce GigaWorld-Policy, an action-centered WAM that learns 2D pixel-action dynamics while enabling efficient action decoding, with optional video generation. Specifically, we formulate policy training into two coupled components: the model predicts future action sequences conditioned on the current observation, and simultaneously generates future videos conditioned on the predicted actions and the same observation. The policy is supervised by both action prediction and video generation, providing richer learning signals and encouraging physically plausible actions through visual-dynamics constraints. With a causal design that prevents future-video tokens from influencing action tokens, explicit future-video generation is optional at inference time, allowing faster action prediction during deployment. To support this paradigm, we curate a diverse, large-scale robot dataset to pre-train an action-centered video generation model, which is then adapted as the backbone for robot policy learning. Experimental results on real-world robotic platforms show that GigaWorld-Policy runs 9x faster than the leading WAM baseline, Motus, while improving task success rates by 7%. Moreover, compared with pi-0.5, GigaWorld-Policy improves performance by 95% on RoboTwin 2.0.

cs.CV↗
arXiv2026

Imagine How To Change: Explicit Procedure Modeling for Change Captioning

Change captioning generates descriptions that explicitly describe the differences between two visually similar images. Existing methods operate on static image pairs, thus ignoring the rich temporal dynamics of the change procedure, which is the key to understand not only what has changed but also how it occurs. We introduce ProCap, a novel framework that reformulates change modeling from static image comparison to dynamic procedure modeling. ProCap features a two-stage design: The first stage trains a procedure encoder to learn the change procedure from a sparse set of keyframes. These keyframes are obtained by automatically generating intermediate frames to make the implicit procedural dynamics explicit and then sampling them to mitigate redundancy. Then the encoder learns to capture the latent dynamics of these keyframes via a caption-conditioned, masked reconstruction task. The second stage integrates this trained encoder within an encoder-decoder model for captioning. Instead of relying on explicit frames from the previous stage -- a process incurring computational overhead and sensitivity to visual noise -- we introduce learnable procedure queries to prompt the encoder for inferring the latent procedure representation, which the decoder then translates into text. The entire model is then trained end-to-end with a captioning loss, ensuring the encoder's output is both temporally coherent and captioning-aligned. Experiments on three datasets demonstrate the effectiveness of ProCap. Code and pre-trained models are available at https://github.com/BlueberryOreo/ProCap

cs.CV↗
arXiv2025

SA-Person: Text-Based Person Retrieval with Scene-aware Re-ranking

Text-based person retrieval aims to identify a target individual from an image gallery using a natural language description. Existing methods primarily focus on appearance-driven cross-modal retrieval, yet face significant challenges due to the visual complexity of scenes and the inherent ambiguity of textual descriptions. The contextual information, such as landmarks and relational cues, provides complementary cues that can offer valuable complementary insights for retrieval, but remains underexploited in current approaches. Motivated by this limitation, we propose a novel paradigm: scene-aware text-based person retrieval, which explicitly integrates both individual appearance and global scene context to improve retrieval accuracy. To support this, we first introduce ScenePerson-13W, a large-scale benchmark dataset comprising over 100,000 real-world scenes with rich annotations encompassing both pedestrian attributes and scene context. Based on this dataset, we further present SA-Person, a two-stage retrieval framework. In the first stage, SA-Person performs discriminative appearance grounding by aligning textual descriptions with pedestrian-specific regions. In the second stage, it introduces SceneRanker, a training-free, scene-aware re-ranking module that refines retrieval results by jointly reasoning over pedestrian appearance and the global scene context. Extensive experiments on ScenePerson-13W and existing benchmarks demonstrate the effectiveness of our proposed SA-Person. Both the dataset and code will be publicly released to facilitate future research.

cs.CV↗
arXiv2025

Text-based Aerial-Ground Person Retrieval

This work introduces Text-based Aerial-Ground Person Retrieval (TAG-PR), which aims to retrieve person images from heterogeneous aerial and ground views with textual descriptions. Unlike traditional Text-based Person Retrieval (T-PR), which focuses solely on ground-view images, TAG-PR introduces greater practical significance and presents unique challenges due to the large viewpoint discrepancy across images. To support this task, we contribute: (1) TAG-PEDES dataset, constructed from public benchmarks with automatically generated textual descriptions, enhanced by a diversified text generation paradigm to ensure robustness under view heterogeneity; and (2) TAG-CLIP, a novel retrieval framework that addresses view heterogeneity through a hierarchically-routed mixture of experts module to learn view-specific and view-agnostic features and a viewpoint decoupling strategy to decouple view-specific features for better cross-modal alignment. We evaluate the effectiveness of TAG-CLIP on both the proposed TAG-PEDES dataset and existing T-PR benchmarks. The dataset and code are available at https://github.com/Flame-Chasers/TAG-PR.

cs.CV↗
arXiv2025

Tuneable ion selectivity in vermiculite membranes intercalated with unexchangeable ions

Membranes selective to ions of the same charge are increasingly sought for wastewater processing and valuable element recovery. However, while narrow channels are known to be essential, other membrane parameters remain difficult to identify and control. Here we show that Zr$^{4+}$, Sn$^{4+}$, Ir$^{4+}$, and La$^{3+}$ ions intercalated into vermiculite laminate membranes become effectively unexchangeable, creating stable channels, one to two water layers wide, that exhibit robust and tuneable ion selectivity. Ion permeability in these membranes spans five orders of magnitude, following a trend dictated by the ions' Gibbs free energy of hydration. Unexpectedly, different intercalated ions lead to two distinct monovalent ion selectivity sequences, despite producing channels of identical width. The selectivity instead correlates with the membranes' stiffness and the entropy of hydration of the intercalated ions. These results introduce a new ion selectivity mechanism driven by entropic and mechanical effects, beyond classical size and charge exclusion.

cond-mat.soft↗
arXiv2025

Multilingual Text-to-Image Person Retrieval via Bidirectional Relation Reasoning and Aligning

Text-to-image person retrieval (TIPR) aims to identify the target person using textual descriptions, facing challenge in modality heterogeneity. Prior works have attempted to address it by developing cross-modal global or local alignment strategies. However, global methods typically overlook fine-grained cross-modal differences, whereas local methods require prior information to explore explicit part alignments. Additionally, current methods are English-centric, restricting their application in multilingual contexts. To alleviate these issues, we pioneer a multilingual TIPR task by developing a multilingual TIPR benchmark, for which we leverage large language models for initial translations and refine them by integrating domain-specific knowledge. Correspondingly, we propose Bi-IRRA: a Bidirectional Implicit Relation Reasoning and Aligning framework to learn alignment across languages and modalities. Within Bi-IRRA, a bidirectional implicit relation reasoning module enables bidirectional prediction of masked image and text, implicitly enhancing the modeling of local relations across languages and modalities, a multi-dimensional global alignment module is integrated to bridge the modality heterogeneity. The proposed method achieves new state-of-the-art results on all multilingual TIPR datasets. Data and code are presented in https://github.com/Flame-Chasers/Bi-IRRA.

cs.CV↗
arXiv2025

Mitigating Hallucinations in Large Vision-Language Models via Entity-Centric Multimodal Preference Optimization

Large Visual Language Models (LVLMs) have demonstrated impressive capabilities across multiple tasks. However, their trustworthiness is often challenged by hallucinations, which can be attributed to the modality misalignment and the inherent hallucinations of their underlying Large Language Models (LLMs) backbone. Existing preference alignment methods focus on aligning model responses with human preferences while neglecting image-text modality alignment, resulting in over-reliance on LLMs and hallucinations. In this paper, we propose Entity-centric Multimodal Preference Optimization (EMPO), which achieves enhanced modality alignment compared to existing human preference alignment methods. Besides, to overcome the scarcity of high-quality multimodal preference data, we utilize open-source instruction datasets to automatically construct high-quality preference data across three aspects: image, instruction, and response. Experiments on two human preference datasets and five multimodal hallucination benchmarks demonstrate the effectiveness of EMPO, e.g., reducing hallucination rates by 85.9\% on Object-HalBench and 49.8\% on MM-HalBench.

cs.CV↗
arXiv2025

Adaptive Sparse Softmax: An Effective and Efficient Softmax Variant

Softmax with the cross entropy loss is the standard configuration for current neural classification models. The gold score for a target class is supposed to be 1, but it is never reachable under the softmax schema. Such a problem makes the training process continue forever and leads to overfitting. Moreover, the "target-approach-1" training goal forces the model to continuously learn all samples, leading to a waste of time in handling some samples which have already been classified correctly with high confidence, while the test goal simply requires the target class of each sample to hold the maximum score. To solve the above weaknesses, we propose the Adaptive Sparse softmax (AS-Softmax) which designs a reasonable and test-matching transformation on top of softmax. For more purposeful learning, we discard the classes with far smaller scores compared with the actual class during training. Then the model could focus on learning to distinguish the target class from its strong opponents, which is also the great challenge in test. In addition, since the training losses of easy samples will gradually drop to 0 in AS-Softmax, we develop an adaptive gradient accumulation strategy based on the masked sample ratio to speed up training. We verify the proposed AS-Softmax on a variety of text multi-class, text multi-label, text token classification, image classification and audio classification tasks with class sizes ranging from 5 to 5000+. The results show that AS-Softmax consistently outperforms softmax and its variants, and the loss of AS-Softmax is remarkably correlated with classification performance in validation. Furthermore, adaptive gradient accumulation strategy can bring about 1.2x training speedup comparing with the standard softmax while maintaining classification effectiveness.

cs.LG↗
arXiv2024

M-Tuning: Prompt Tuning with Mitigated Label Bias in Open-Set Scenarios

In realistic open-set scenarios where labels of a part of testing data are totally unknown, when vision-language (VL) prompt learning methods encounter inputs related to unknown classes (i.e., not seen during training), they always predict them as one of the training classes. The exhibited label bias causes difficulty in open set recognition (OSR), in which an image should be correctly predicted as one of the known classes or the unknown one. To achieve this goal, we propose a vision-language prompt tuning method with mitigated label bias (M-Tuning). It introduces open words from the WordNet to extend the range of words forming the prompt texts from only closed-set label words to more, and thus prompts are tuned in a simulated open-set scenario. Besides, inspired by the observation that classifying directly on large datasets causes a much higher false positive rate than on small datasets, we propose a Combinatorial Tuning and Testing (CTT) strategy for improving performance. CTT decomposes M-Tuning on large datasets as multiple independent group-wise tuning on fewer classes, then makes accurate and comprehensive predictions by selecting the optimal sub-prompt. Finally, given the lack of VL-based OSR baselines in the literature, especially for prompt methods, we contribute new baselines for fair comparisons. Our method achieves the best performance on datasets with various scales, and extensive ablation studies also validate its effectiveness.

cs.CV↗
arXiv2024

MAIR: A Massive Benchmark for Evaluating Instructed Retrieval

Recent information retrieval (IR) models are pre-trained and instruction-tuned on massive datasets and tasks, enabling them to perform well on a wide range of tasks and potentially generalize to unseen tasks with instructions. However, existing IR benchmarks focus on a limited scope of tasks, making them insufficient for evaluating the latest IR models. In this paper, we propose MAIR (Massive Instructed Retrieval Benchmark), a heterogeneous IR benchmark that includes 126 distinct IR tasks across 6 domains, collected from existing datasets. We benchmark state-of-the-art instruction-tuned text embedding models and re-ranking models. Our experiments reveal that instruction-tuned models generally achieve superior performance compared to non-instruction-tuned models on MAIR. Additionally, our results suggest that current instruction-tuned text embedding models and re-ranking models still lack effectiveness in specific long-tail tasks. MAIR is publicly available at https://github.com/sunnweiwei/Mair.

cs.IR↗