arXiv subjects

Haijun Li

Publications and source records attributed to Haijun Li.

At least 19 recordsLinked to original sources

arXiv2026

OvisOCR2 Technical Report

We introduce OvisOCR2, a 0.8B document parsing model. OvisOCR2 is designed as an end-to-end parser: given a document page image, it generates a Markdown representation in natural reading order, covering text, formulas, tables, and visual regions. We build a data engine that combines filtered real-document annotations with synthetic pages whose rendered images and Markdown targets are derived from the same HTML source. The training recipe includes supervised fine-tuning, reinforcement learning on a 4B branch with a multi-component reward design, on-policy distillation into the 0.8B model, and model fusion. On OmniDocBench v1.6, OvisOCR2 achieves a state-of-the-art overall score of 96.58, placing an end-to-end model at the top of this leaderboard previously dominated by pipeline methods and highlighting the potential of end-to-end document parsing. On PureDocBench, OvisOCR2 also achieves the highest Avg3 score of 75.06. Beyond these two public benchmarks, we evaluate OvisOCR2 on an in-house benchmark designed to cover a broader set of long-tail and challenging scenarios. OvisOCR2 obtains the best overall performance among the compared methods, providing further evidence of its generalization and robustness. OvisOCR2 is available at https://huggingface.co/ATH-MaaS/OvisOCR2.

cs.CV↗

arXiv2026

A Task-Driven Evaluation of UAV Detection and Tracking under Synthetic Fog

Fog severely degrades the visibility of small unmanned aerial vehicles (UAVs) in skydominant, long-range imagery, reducing the reliability of downstream detection and tracking. This paper presents a task-driven evaluation framework that links depth-aware synthetic fog generation, image restoration, object detection, and tracking within a unified pipeline. Given the practical difficulty of collecting and annotating foggy UAV scenes, synthetic fog is generated from real clear-weather outdoor images containing UAV targets using monocular depth estimation and the atmospheric scattering model. Representative restoration methods from classical, convolutional neural network (CNN)-based, and transformer-based families are first compared, after which the selected restoration model is integrated into the downstream perception pipeline. Detection is evaluated under both clean-only and fog-inclusive training regimes using multiple detector variants, while tracking-by-detection is assessed on clean, foggy, and restored video sequences. Beyond image-level restoration metrics, the study evaluates how fog and restoration affect detection robustness and tracking performance. The results show that fog substantially degrades both detection and tracking, primarily through increased missed detections. Fog-inclusive training provides the most consistent improvement in robustness, whereas test-time restoration is most beneficial when the detector has been trained only on clean imagery. These findings show that restoration quality does not necessarily translate into proportional gains in downstream perception and therefore should be evaluated jointly with detection and tracking performance.

cs.CV↗

arXiv2026

Proofs of five conjectural identities on modular rank four Nahm sums

Nahm sums and Rogers-Ramanujan type identities have attracted considerable attention in recent years. In this paper, we provide analytic proofs of five conjectural identities on modular rank four Nahm sums that were proposed by Cao and Wang. Moreover, we reveal that the conjectures of Shi-Wang and Cao-Wang are closely related.

math.NT↗

arXiv2026

Linked partition ideals and gap-frequency partitions

Recently, linked partition ideals have attracted renewed attention. In this paper, we embed the gap-frequency partitions in which each part appears at most twice or three times into the framework of span one linked partition ideals, and derive refined generating functions for both cases. Furthermore, for gap-frequency partitions in which every part appears at most k times, we provide a purely combinatorial proof of the corresponding refined generating function.

math.CO↗

arXiv2026

On mesh patterns of short length: Equidistribution and enumeration

The classification and enumeration of short mesh patterns have emerged as two central directions in the area. We make substantial progress on both fronts. We construct an involution and a bijection that establish distributional equivalences for two classes of length-$2$ mesh patterns, thereby resolving a conjecture from 2019 and a recent conjecture. As a consequence, the best known upper bounds for the numbers of distribution-equivalence and Wilf-equivalence classes drop to $106$ and $47$, respectively. Combined with the known lower bounds of 105 and 46, conjectured to be exact, these results leave both classifications hinging on a single distribution-equivalence question conjectured in 2019, whose resolution would at once settle the remaining Wilf-equivalence case. We further conjecture that this unresolved equidistribution also holds for involutions, a subclass of all permutations. We also determine the distributions of three additional classes of length-$2$ mesh patterns through a detailed structural analysis. Our work combines bijective techniques with generating-function methods, yielding new insights into the structure and enumeration of short mesh patterns.

math.CO↗

arXiv2026

Combinatorial perspectives on identities for partitions with distinct even parts

Partitions with distinct even parts have long been the subject of extensive research. In this paper, We present some new perspectives on such partitions from a combinatorial viewpoint, and connect them with signed partitions and bicolored partitions, thereby obtaining several partition identities. We construct bijective proofs for each of our results. Furthermore, these bijections will partially answer the combinatorial problems posed by Andrews-El Bachraoui and K$\imath$l$\imath$\c{c}-Kur\c{s}ung\"oz. respectively.

math.CO↗

arXiv2026

Table-as-Search: Formulate Long-Horizon Agentic Information Seeking as Table Completion

Current Information Seeking (InfoSeeking) agents struggle to maintain focus and coherence during long-horizon exploration, as tracking search states, including planning procedure and massive search results, within one plain-text context is inherently fragile. To address this, we introduce \textbf{Table-as-Search (TaS)}, a structured planning framework that reformulates the InfoSeeking task as a Table Completion task. TaS maps each query into a structured table schema maintained in an external database, where rows represent search candidates and columns denote constraints or required information. This table precisely manages the search states: filled cells strictly record the history and search results, while empty cells serve as an explicit search plan. Crucially, TaS unifies three distinct InfoSeeking tasks: Deep Search, Wide Search, and the challenging DeepWide Search. Extensive experiments demonstrate that TaS significantly outperforms numerous state-of-the-art baselines across three kinds of benchmarks, including multi-agent framework and commercial systems. Furthermore, our analysis validates the TaS's superior robustness in long-horizon InfoSeeking, alongside its efficiency, scalability and flexibility. Code and datasets are publicly released at https://github.com/AIDC-AI/Marco-Search-Agent.

cs.CL↗

arXiv2026

Vectra: A New Metric, Dataset, and Model for Visual Quality Assessment in E-Commerce In-Image Machine Translation

In-Image Machine Translation (IIMT) powers cross-border e-commerce product listings; existing research focuses on machine translation evaluation, while visual rendering quality is critical for user engagement. When facing context-dense product imagery and multimodal defects, current reference-based methods (e.g., SSIM, FID) lack explainability, while model-as-judge approaches lack domain-grounded, fine-grained reward signals. To bridge this gap, we introduce Vectra, to the best of our knowledge, the first reference-free, MLLM-driven visual quality assessment framework for e-commerce IIMT. Vectra comprises three components: (1) Vectra Score, a multidimensional quality metric system that decomposes visual quality into 14 interpretable dimensions, with spatially-aware Defect Area Ratio (DAR) quantification to reduce annotation ambiguity; (2) Vectra Dataset, constructed from 1.1M real-world product images via diversity-aware sampling, comprising a 2K benchmark for system evaluation, 30K reasoning-based annotations for instruction tuning, and 3.5K expert-labeled preferences for alignment and evaluation; and (3) Vectra Model, a 4B-parameter MLLM that generates both quantitative scores and diagnostic reasoning. Experiments demonstrate that Vectra achieves state-of-the-art correlation with human rankings, and our model outperforms leading MLLMs, including GPT-5 and Gemini-3, in scoring performance. The dataset and model will be released upon acceptance.

cs.CV↗

arXiv2026

Combinatorial proofs of some identities on overpartitions with repeated smallest non-overlined part

Let $\overline{\mathrm{spt}}k(n)$ denote the number of overpartitions of $n$ where the smallest non-overlined part, say $s(\pi)$, appears $k$ times and every overlined part is bigger than $s(\pi)$. Let $\overline{\mathrm{spt}}k_o(n)$ denote the number of overpartitions of $n$ where the smallest non-overlined part appears $k$ times, every overlined part is bigger than $s(\pi)$ and all parts other than $s(\pi)$ are incongruent modulo $2$ with $s(\pi)$. Also, let $b_e(k,n)$ (resp., $b_o(k,n)$) denote the number of overpartitions of $n$ counted by $\overline{\mathrm{spt}}k_o(n)$ where the number of parts greater than $s(\pi)$ is even (resp., odd), and let $$\overline{\mathrm{spt}}k_o'(n)=b_e(k,n)-b_o(k,n).$$ Recently, Malik and Sarma (arXiv:2601.15601v1) expressed the generating functions of these partition functions in terms of linear combinations of $q$-series with polynomials in $q$ as coefficients. As corollaries, they derived some partition identities involving the functions for $k=1$ and sought for combinatorial proofs of their results. In this paper, we present some desired proofs.

math.CO↗

arXiv2025

Operator Tail Densities of Multivariate Copulas

Operator regular variation of a multivariate distribution can be decomposed into the operator tail dependence of the underlying copula and the regular variation of the univariate marginals. In this paper, we introduce operator tail densities for copulas and show that an operator-regularly-varying density can be characterized through the operator tail density of its copula together with the marginal regular variation. As an example, we demonstrate that although a Liouville copula is not available in closed form, it nevertheless admits an explicit operator tail-dependence function.

math.ST↗

arXiv2025

DeepWideSearch: Benchmarking Depth and Width in Agentic Information Seeking

Current search agents fundamentally lack the ability to simultaneously perform \textit{deep} reasoning over multi-hop retrieval and \textit{wide}-scale information collection-a critical deficiency for real-world applications like comprehensive market analysis and business development. To bridge this gap, we introduce DeepWideSearch, the first benchmark explicitly designed to evaluate agents to integrate depth and width in information seeking. In DeepWideSearch, agents must process a large volume of data, each requiring deep reasoning over multi-hop retrieval paths. Specifically, we propose two methods to converse established datasets, resulting in a curated collection of 220 questions spanning 15 diverse domains. Extensive experiments demonstrate that even state-of-the-art agents achieve only 2.39% average success rate on DeepWideSearch, highlighting the substantial challenge of integrating depth and width search in information-seeking tasks. Furthermore, our error analysis reveals four failure modes: lack of reflection, overreliance on internal knowledge, insufficient retrieval, and context overflow-exposing key limitations in current agent architectures. We publicly release DeepWideSearch to catalyze future research on more capable and robust information-seeking agents.

cs.CL↗

arXiv2025

Parity patterns meet Genocchi numbers, I: four labelings and three bijections

Hetyei introduced in 2019 the homogenized Linial arrangement and showed that its regions are counted by the median Genocchi numbers. In the course of devising a different proof of Hetyei's result, Lazar and Wachs considered another hyperplane arrangement that is associated with certain bipartite graph called Ferrers graph. We bijectively label the regions of this latter arrangement with permutations whose ascents are subject to a parity restriction. This labeling not only establishes the equivalence between two enumerative results due to Hetyei and Lazar-Wachs, repectively, but also motivates us to derive and investigate a Seidel-like triangle that interweaves Genocchi numbers of both kinds. Applying similar ideas, we introduce three more variants of permutations with analogous parity restrictions. We provide labelings for regions of the aforementioned arrangement using these three sets of restricted permutations as well. Furthermore, bijections from our first permutation model to two previously known permutation models are established.

math.CO↗

arXiv2025

Sequences of odd length in strict partitions IV: the combinatorics of parameterized Rogers-Ramanujan type identities

In the first three papers, we conducted a series of discussions on the statistics of strict partitions and Rogers-Ramanujan partitions, specifically the sequences of odd length (denoted as $\mathrm{sol}$) and its extensions. We established bijections for some Rogers-Ramanujan type identities. This paper will continue that series of work, and first we will use the bijective method to re-establish several parameterized Rogers-Ramanujan type identities, which appeared in the recent work of Hao-Kuai-Xia and Li-Wang. Moreover, we focus on the work of Chen-Yin and parameterize their main results, where the $\mathrm{sol}$ has evolved.

math.CO↗

arXiv2025

Ovis2.5 Technical Report

We present Ovis2.5, a successor to Ovis2 designed for native-resolution visual perception and strong multimodal reasoning. Ovis2.5 integrates a native-resolution vision transformer that processes images at their native, variable resolutions, avoiding the degradation from fixed-resolution tiling and preserving both fine detail and global layout -- crucial for visually dense content like complex charts. To strengthen reasoning, we train the model to move beyond linear chain-of-thought and perform reflection -- including self-checking and revision. This advanced capability is exposed as an optional "thinking mode" at inference time, allowing users to trade latency for enhanced accuracy on difficult inputs. The model is trained via a comprehensive five-phase curriculum that progressively builds its skills. The process begins with foundational visual and multimodal pretraining, advances through large-scale instruction tuning, and culminates in alignment and reasoning enhancement using DPO and GRPO. To scale these upgrades efficiently, we employ multimodal data packing and hybrid parallelism, yielding a significant end-to-end speedup. We release two open-source models: Ovis2.5-9B and Ovis2.5-2B. The latter continues the "small model, big performance" philosophy of Ovis2, making it ideal for resource-constrained, on-device scenarios. On the OpenCompass multimodal leaderboard, Ovis2.5-9B averages 78.3, marking a substantial improvement over its predecessor, Ovis2-8B, and achieving state-of-the-art results among open-source MLLMs in the sub-40B parameter range; Ovis2.5-2B scores 73.9, establishing SOTA for its size. Beyond aggregate scores, Ovis2.5 achieves leading results on STEM benchmarks, exhibits strong capabilities on grounding and video tasks, and achieves open-source SOTA at its scale for complex chart analysis.

cs.CV↗

arXiv2025

Marco-Voice Technical Report

This paper presents a multifunctional speech synthesis system that integrates voice cloning and emotion control speech synthesis within a unified framework. The goal of this work is to address longstanding challenges in achieving highly expressive, controllable, and natural speech generation that faithfully preserves speaker identity across diverse linguistic and emotional contexts. Our approach introduces an effective speaker-emotion disentanglement mechanism with in-batch contrastive learning, enabling independent manipulation of speaker identity and eemotional style, as well as rotational emotional embedding integration method for smooth emotion control. To support comprehensive training and evaluation, we construct CSEMOTIONS, a high-quality emotional speech dataset containing 10 hours of Mandarin speech from ten professional speakers across seven emotional categories. Extensive experiments demonstrate that our system, Marco-Voice, achieves substantial improvements in both objective and subjective metrics. Comprehensive evaluations and analysis were conducted, results show that MarcoVoice delivers competitive performance in terms of speech clarity and emotional richness, representing a substantial advance in the field of expressive neural speech synthesis. Our code and dataset are publicly available at https://github.com/AIDC-AI/Marco-Voice and https://huggingface.co/datasets/AIDC-AI/CSEMOTIONS respectively.

cs.CL↗

arXiv2025

On fourteen equidistribution conjectures of Lv and Zhang and monotone mesh patterns with corner shadings

Three complementation-like involutions are constructed on permutations to prove, and in some cases generalize, all remaining fourteen joint symmetric equidistribution conjectures of Lv and Zhang. Further enumerative results are obtained for several classes of (mesh) pattern-avoiding permutations, where the shadings of all involved mesh patterns are restricted to an opposing pair of corners.

math.CO↗

arXiv2025

The combinatorics of identities involving overpartitions with distinct parts

Recently, Andrews and EI Bachraoui discovered several companions for some famous $q$-series formulas, and derived some new identities involving partitions and overpartitions with distinct parts. In this paper, we shall refine their results by the number of parts of partitions and furthermore, we will also provide the combinatorial proofs for those partition identities.

math.CO↗

arXiv2025

TransBench: Benchmarking Machine Translation for Industrial-Scale Applications

Machine translation (MT) has become indispensable for cross-border communication in globalized industries like e-commerce, finance, and legal services, with recent advancements in large language models (LLMs) significantly enhancing translation quality. However, applying general-purpose MT models to industrial scenarios reveals critical limitations due to domain-specific terminology, cultural nuances, and stylistic conventions absent in generic benchmarks. Existing evaluation frameworks inadequately assess performance in specialized contexts, creating a gap between academic benchmarks and real-world efficacy. To address this, we propose a three-level translation capability framework: (1) Basic Linguistic Competence, (2) Domain-Specific Proficiency, and (3) Cultural Adaptation, emphasizing the need for holistic evaluation across these dimensions. We introduce TransBench, a benchmark tailored for industrial MT, initially targeting international e-commerce with 17,000 professionally translated sentences spanning 4 main scenarios and 33 language pairs. TransBench integrates traditional metrics (BLEU, TER) with Marco-MOS, a domain-specific evaluation model, and provides guidelines for reproducible benchmark construction. Our contributions include: (1) a structured framework for industrial MT evaluation, (2) the first publicly available benchmark for e-commerce translation, (3) novel metrics probing multi-level translation quality, and (4) open-sourced evaluation tools. This work bridges the evaluation gap, enabling researchers and practitioners to systematically assess and enhance MT systems for industry-specific needs.

cs.CL↗