arXiv subjects

Shihao Chen

Publications and source records attributed to Shihao Chen.

At least 19 recordsLinked to original sources

arXiv2026

PixelEyes: Decoupling Perception and Reasoning for Pinpoint Visual Evidence Seeking

This paper explores multi-turn visual reasoning and observes that MLLMs repeatedly fail to localize the target, leading to long, redundant trajectories. We attribute this failure to the entanglement of reasoning and perception within a single model, the MLLM reasons and localizes simultaneously, and inaccurate localization triggers additional reasoning turns that bloat the trajectory. To solve this problem, we propose PixelEyes, a multi-turn visual reasoning agent that explicitly decouples reasoning from perception, i.e., the reasoner decides what to look for, while a specialized perception tool answers where it is. Specifically, PixelEyes introduces 1) Mask-guided Visual Search. A referring segmentation model is invoked to provide mask-precise localization, freeing the reasoner from the need to compensate for imprecise grounding. 2) Semantic-region Breadth-first Search (BFS). To eliminate redundant loops caused by repeatedly cropping incorrect sub-regions, we organize exploration as a breadth-first search over semantic regions. To internalize these capabilities, we construct the PixelEyes-6K dataset by resynthesizing expert trajectories from existing data. This explicitly embeds our mask-guided search and BFS logic into the model. We further introduce Pinpoint-Bench, a zero-hint visual search benchmark, i.e., no location cues are provided in the question, with instance-level masks and bounding boxes that separate localization failures from reasoning failures, enabling fine-grained analysis of failure modes such as inattentional blindness. Recent state-of-the-art MLLMs and visual reasoning agents leave large headroom on Pinpoint-Bench, demonstrating its quality and difficulty. Code and models are open-sourced.

cs.CV↗

arXiv2026

Towards One-to-Many Temporal Grounding

Temporal Grounding (TG) aims to localize video segments corresponding to a textual query. Prior research predominantly focuses on single-segment retrieval. Real-world scenarios, however, often require localizing multiple disjoint segments for a single query -- a setting we term One-to-Many Temporal Grounding (OMTG). Previous state-of-the-art MLLMs, optimized for one-to-one settings, struggle in this context, often yielding near-zero scores due to a lack of event cardinality perception. To bridge this gap, we present a systematic solution with three key contributions. First, we establish the first comprehensive OMTG benchmark, introducing Count Accuracy (C-Acc) and Effective Temporal F1 (EtF1) as evaluation metrics. Second, we curate a high-quality OMTG dataset comprising 56k samples through a sophisticated construction pipeline. Third, we develop novel temporal and caption reward functions specifically designed for OMTG. In particular, the caption reward leverages Chain-of-Thought reasoning over dense video captions to explicitly guide policy optimization toward both preciseness and completeness. Extensive experiments show our model achieves a new state-of-the-art EtF1 of 43.65\% on OMTG Bench, outperforming Gemini 2.5 Pro and Seed-1.8 by 15.85\% and 15.61\%, respectively. Project Page: https://insomniaaac.github.io/OMTG/

cs.CV↗

arXiv2026

Report of the 5th PVUW Challenge: Towards More Diverse Modalities in Pixel-Level Understanding

This report summarizes the objectives, datasets, and top-performing methodologies of the 2026 Pixel-level Video Understanding in the Wild (PVUW) Challenge, hosted at CVPR 2026, which evaluates state-of-the-art models under highly unconstrained conditions. To provide a comprehensive assessment, the 2026 edition features three specialized tracks: the MOSE track for tracking objects within densely cluttered and severely occluded scenarios; the MeViS-Text track for localizing targets via motion-focused linguistic expressions; and the newly inaugurated MeViS-Audio track, which pioneers acoustic-driven object segmentation. By introducing previously unreleased challenging data and analyzing the cutting-edge, multimodal solutions submitted by participants, this report highlights the community's latest technical advancements and charts promising future directions for robust video scene comprehension.

cs.CV↗

arXiv2026

HAFM: Hierarchical Autoregressive Foundation Model for Music Accompaniment Generation

Music accompaniment generation aims to automatically produce instrumental accompaniments that are rhythmically, harmonically, and timbrally coherent with a given vocal input, with broad applications in personalized music creation, arrangement assistance, and music education. Existing approaches, primarily operating in the symbolic domain or relying on single-stage audio generation frameworks, commonly suffer from insufficient high-level semantic structure modeling, limited acoustic detail reconstruction, and weak conditional controllability. To address these limitations, this paper proposes HAFM, a Hierarchical Autoregressive Foundation Model for vocal-conditioned music accompaniment generation. The model employs a dual-rate tokenization strategy in which $50$ Hz HuBERT semantic tokens capture high-level musical structure and $75$ Hz EnCodec acoustic tokens encode fine-grained acoustic content, enabling explicit disentanglement of semantic and acoustic representations. Building on this foundation, a three-stage cascaded generation framework is designed to progressively generate semantic tokens, coarse acoustic tokens, and fine acoustic tokens, refining the accompaniment from global structure to local detail. . Objective evaluation on the MUSDB18 dataset demonstrates that the full three-stage model achieves a Fr{\'e}chet Audio Distance (FAD) score of 1.71, representing an 18.6% relative improvement over the two-stage baseline (FAD = 2.10). Subjective listening tests show that the generated accompaniments achieve a 51.5% preference rate against ground-truth accompaniments in head-to-head comparisons, and substantially outperform the random baseline in terms of rhythmic alignment, harmonic compatibility, and overall musical coherence. The source code and demo are available at https://github.com/HackerHyper/HAFM.git.

cs.SD↗

arXiv2026

SaSaSaSa2VA: 2nd Place of the 5th PVUW MeViS-Text Track

Referring video object segmentation (RVOS) commonly grounds targets in videos based on static textual cues. MeViS benchmark extends this by incorporating motion-centric expressions (referring & reasoning motion expressions) and introducing no-target queries. Extending SaSaSa2VA, where increased input frames and [SEG] tokens already strengthen the Sa2VA backbone, we adopt a simple yet effective target existence-aware verification mechanism, leading to Still Awesome SaSaSa2VA (SaSaSaSa2VA). Despite its simplicity, the method achieves a final score of 89.19 in the 5th PVUW Challenge (MeViS-Text Track), securing 2nd place. Both quantitative results and ablations suggest that this existence-aware verification strategy is sufficient to unlock strong performance on motion-centric referring tasks.

cs.CV↗

arXiv2025

LSVOS 2025 Challenge Report: Recent Advances in Complex Video Object Segmentation

This report presents an overview of the 7th Large-scale Video Object Segmentation (LSVOS) Challenge held in conjunction with ICCV 2025. Besides the two traditional tracks of LSVOS that jointly target robustness in realistic video scenarios: Classic VOS (VOS), and Referring VOS (RVOS), the 2025 edition features a newly introduced track, Complex VOS (MOSEv2). Building upon prior insights, MOSEv2 substantially increases difficulty, introducing more challenging but realistic scenarios including denser small objects, frequent disappear/reappear events, severe occlusions, adverse weather and lighting, etc., pushing long-term consistency and generalization beyond curated benchmarks. The challenge retains standard ${J}$, $F$, and ${J\&F}$ metrics for VOS and RVOS, while MOSEv2 adopts ${J\&\dot{F}}$ as the primary ranking metric to better evaluate objects across scales and disappearance cases. We summarize datasets and protocols, highlight top-performing solutions, and distill emerging trends, such as the growing role of LLM/MLLM components and memory-aware propagation, aiming to chart future directions for resilient, language-aware video segmentation in the wild.

cs.CV↗

arXiv2025

The 1st Solution for 7th LSVOS RVOS Track: SaSaSa2VA

Referring video object segmentation (RVOS) requires segmenting and tracking objects in videos conditioned on natural-language expressions, demanding fine-grained understanding of both appearance and motion. Building on Sa2VA, which couples a Multi-modal Large Language Model (MLLM) with the video segmentation model SAM2, we identify two key bottlenecks that limit segmentation performance: sparse frame sampling and reliance on a single [SEG] token for an entire video. We propose Segmentation Augmented and Selective Averaged Sa2VA (SaSaSa2VA) to address these issues. On the 7th LSVOS Challenge (RVOS track), SaSaSa2VA achieves a $\mathcal{J\&F}$ of 67.45, ranking first and surpassing the runner-up by 2.80 points. This result and ablation studies demonstrate that efficient segmentation augmentation and test-time ensembling substantially enhance grounded MLLMs for RVOS. The code is released in Sa2VA repository: https://github.com/bytedance/Sa2VA.

cs.CV↗

arXiv2025

Beyond Appearance: Geometric Cues for Robust Video Instance Segmentation

Video Instance Segmentation (VIS) fundamentally struggles with pervasive challenges including object occlusions, motion blur, and appearance variations during temporal association. To overcome these limitations, this work introduces geometric awareness to enhance VIS robustness by strategically leveraging monocular depth estimation. We systematically investigate three distinct integration paradigms. Expanding Depth Channel (EDC) method concatenates the depth map as input channel to segmentation networks; Sharing ViT (SV) designs a uniform ViT backbone, shared between depth estimation and segmentation branches; Depth Supervision (DS) makes use of depth prediction as an auxiliary training guide for feature learning. Though DS exhibits limited effectiveness, benchmark evaluations demonstrate that EDC and SV significantly enhance the robustness of VIS. When with Swin-L backbone, our EDC method gets 56.2 AP, which sets a new state-of-the-art result on OVIS benchmark. This work conclusively establishes depth cues as critical enablers for robust video understanding.

cs.CV↗

arXiv2025

VectorLLM: Human-like Extraction of Structured Building Contours vis Multimodal LLMs

Automatically extracting vectorized building contours from remote sensing imagery is crucial for urban planning, population estimation, and disaster assessment. Current state-of-the-art methods rely on complex multi-stage pipelines involving pixel segmentation, vectorization, and polygon refinement, which limits their scalability and real-world applicability. Inspired by the remarkable reasoning capabilities of Large Language Models (LLMs), we introduce VectorLLM, the first Multi-modal Large Language Model (MLLM) designed for regular building contour extraction from remote sensing images. Unlike existing approaches, VectorLLM performs corner-point by corner-point regression of building contours directly, mimicking human annotators' labeling process. Our architecture consists of a vision foundation backbone, an MLP connector, and an LLM, enhanced with learnable position embeddings to improve spatial understanding capability. Through comprehensive exploration of training strategies including pretraining, supervised fine-tuning, and preference optimization across WHU, WHU-Mix, and CrowdAI datasets, VectorLLM significantly outperformed the previous SOTA methods by 5.6 AP, 7.1 AP, 13.6 AP, respectively in the three datasets. Remarkably, VectorLLM exhibits strong zero-shot performance on unseen objects including aircraft, water bodies, and oil tanks, highlighting its potential for unified modeling of diverse remote sensing object contour extraction tasks. Overall, this work establishes a new paradigm for vector extraction in remote sensing, leveraging the topological reasoning capabilities of LLMs to achieve both high accuracy and exceptional generalization. All the codes and weights will be published for promoting community development.

cs.CV↗

arXiv2025

DenseWorld-1M: Towards Detailed Dense Grounded Caption in the Real World

Multimodal Large Language Models (MLLMs) demonstrate a complex understanding of scenes, benefiting from large-scale and high-quality datasets. Most existing caption datasets lack the ground locations and relations for visual entities. Several grounded caption datasets face the problems of missing detailed descriptions, relations, and massive object descriptions on high-resolution images. To fill this gap for the community, we present DenseWorld-1M, the first massive, detailed, dense grounded caption dataset in the real world. We design a three-stage labeling pipeline, containing open-world perception, detailed object caption generation, and dense caption merging. The first stage obtains entity-level masks and labels. The second stage generates the object-level, detailed captions with the guidance of masks and labels from the first stage. The final stage merges object captions and masks into spatial and relational dense captions. To accelerate the labeling process and improve caption quality, we present two VLM models: the Detailed Region Caption model and the Spatial Caption Merging model. Extensive experiments on various settings, including vision-language understanding, visual grounding, and region caption generation, demonstrate the effectiveness of our DenseWorld-1M dataset and labeling models.

cs.CV↗

arXiv2025

Pixel-SAIL: Single Transformer For Pixel-Grounded Understanding

Multimodal Large Language Models (MLLMs) achieve remarkable performance for fine-grained pixel-level understanding tasks. However, all the works rely heavily on extra components, such as vision encoder (CLIP), segmentation experts, leading to high system complexity and limiting model scaling. In this work, our goal is to explore a highly simplified MLLM without introducing extra components. Our work is motivated by the recent works on Single trAnsformer as a unified vIsion-Language Model (SAIL) design, where these works jointly learn vision tokens and text tokens in transformers. We present Pixel-SAIL, a single transformer for pixel-wise MLLM tasks. In particular, we present three technical improvements on the plain baseline. First, we design a learnable upsampling module to refine visual token features. Secondly, we propose a novel visual prompt injection strategy to enable the single transformer to understand visual prompt inputs and benefit from the early fusion of visual prompt embeddings and vision tokens. Thirdly, we introduce a vision expert distillation strategy to efficiently enhance the single transformer's fine-grained feature extraction capability. In addition, we have collected a comprehensive pixel understanding benchmark (PerBench), using a manual check. It includes three tasks: detailed object description, visual prompt-based question answering, and visual-text referring segmentation. Extensive experiments on four referring segmentation benchmarks, one visual prompt benchmark, and our PerBench show that our Pixel-SAIL achieves comparable or even better results with a much simpler pipeline. Code and model will be released at https://github.com/magic-research/Sa2VA.

cs.CV↗

arXiv2025

Are They the Same? Exploring Visual Correspondence Shortcomings of Multimodal LLMs

Recent advancements in multimodal large language models (MLLM) have shown a strong ability in visual perception, reasoning abilities, and vision-language understanding. However, the visual matching ability of MLLMs is rarely studied, despite finding the visual correspondence of objects is essential in computer vision. Our research reveals that the matching capabilities in recent MLLMs still exhibit systematic shortcomings, even with current strong MLLMs models, GPT-4o. In particular, we construct a Multimodal Visual Matching (MMVM) benchmark to fairly benchmark over 30 different MLLMs. The MMVM benchmark is built from 15 open-source datasets and Internet videos with manual annotation. We categorize the data samples of MMVM benchmark into eight aspects based on the required cues and capabilities to more comprehensively evaluate and analyze current MLLMs. In addition, we have designed an automatic annotation pipeline to generate the MMVM SFT dataset, including 220K visual matching data with reasoning annotation. To our knowledge, this is the first visual corresponding dataset and benchmark for the MLLM community. Finally, we present CoLVA, a novel contrastive MLLM with two novel technical designs: fine-grained vision expert with object-level contrastive learning and instruction augmentation strategy. The former learns instance discriminative tokens, while the latter further improves instruction following ability. CoLVA-InternVL2-4B achieves an overall accuracy (OA) of 49.80\% on the MMVM benchmark, surpassing GPT-4o and the best open-source MLLM, Qwen2VL-72B, by 7.15\% and 11.72\% OA, respectively. These results demonstrate the effectiveness of our MMVM SFT dataset and our novel technical designs. Code, benchmark, dataset, and models will be released.

cs.CV↗

arXiv2024

CSSinger: End-to-End Chunkwise Streaming Singing Voice Synthesis System Based on Conditional Variational Autoencoder

Singing Voice Synthesis (SVS) aims to generate singing voices of high fidelity and expressiveness. Conventional SVS systems usually utilize an acoustic model to transform a music score into acoustic features, followed by a vocoder to reconstruct the singing voice. It was recently shown that end-to-end modeling is effective in the fields of SVS and Text to Speech (TTS). In this work, we thus present a fully end-to-end SVS method together with a chunkwise streaming inference to address the latency issue for practical usages. Note that this is the first attempt to fully implement end-to-end streaming audio synthesis using latent representations in VAE. We have made specific improvements to enhance the performance of streaming SVS using latent representations. Experimental results demonstrate that the proposed method achieves synthesized audio with high expressiveness and pitch accuracy in both streaming SVS and TTS tasks.

eess.AS↗

arXiv2024

A Lightweight and Real-Time Binaural Speech Enhancement Model with Spatial Cues Preservation

Binaural speech enhancement (BSE) aims to jointly improve the speech quality and intelligibility of noisy signals received by hearing devices and preserve the spatial cues of the target for natural listening. Existing methods often suffer from the compromise between noise reduction (NR) capacity and spatial cues preservation (SCP) accuracy and a high computational demand in complex acoustic scenes. In this work, we present a learning-based lightweight binaural complex convolutional network (LBCCN), which excels in NR by filtering low-frequency bands and keeping the rest. Additionally, our approach explicitly incorporates the estimation of interchannel relative acoustic transfer function to ensure the spatial cues fidelity and speech clarity. Results show that the proposed LBCCN can achieve a comparable NR performance to state-of-the-art methods under fixed-speaker conditions, but with a much lower computational cost and a certain degree of SCP capability. The reproducible code and audio examples are available at https://github.com/jywanng/LBCCN.

cs.SD↗

arXiv2024

LCM-SVC: Latent Diffusion Model Based Singing Voice Conversion with Inference Acceleration via Latent Consistency Distillation

Any-to-any singing voice conversion (SVC) aims to transfer a target singer's timbre to other songs using a short voice sample. However many diffusion model based any-to-any SVC methods, which have achieved impressive results, usually suffered from low efficiency caused by a mass of inference steps. In this paper, we propose LCM-SVC, a latent consistency distillation (LCD) based latent diffusion model (LDM) to accelerate inference speed. We achieved one-step or few-step inference while maintaining the high performance by distilling a pre-trained LDM based SVC model, which had the advantages of timbre decoupling and sound quality. Experimental results show that our proposed method can significantly reduce the inference time and largely preserve the sound quality and timbre similarity comparing with other state-of-the-art SVC models. Audio samples are available at https://sounddemos.github.io/lcm-svc.

eess.AS↗

arXiv2024

LDM-SVC: Latent Diffusion Model Based Zero-Shot Any-to-Any Singing Voice Conversion with Singer Guidance

Any-to-any singing voice conversion (SVC) is an interesting audio editing technique, aiming to convert the singing voice of one singer into that of another, given only a few seconds of singing data. However, during the conversion process, the issue of timbre leakage is inevitable: the converted singing voice still sounds like the original singer's voice. To tackle this, we propose a latent diffusion model for SVC (LDM-SVC) in this work, which attempts to perform SVC in the latent space using an LDM. We pretrain a variational autoencoder structure using the noted open-source So-VITS-SVC project based on the VITS framework, which is then used for the LDM training. Besides, we propose a singer guidance training method based on classifier-free guidance to further suppress the timbre of the original singer. Experimental results show the superiority of the proposed method over previous works in both subjective and objective evaluations of timbre similarity.

eess.AS↗

arXiv2024

A Study of Dropout-Induced Modality Bias on Robustness to Missing Video Frames for Audio-Visual Speech Recognition

Advanced Audio-Visual Speech Recognition (AVSR) systems have been observed to be sensitive to missing video frames, performing even worse than single-modality models. While applying the dropout technique to the video modality enhances robustness to missing frames, it simultaneously results in a performance loss when dealing with complete data input. In this paper, we investigate this contrasting phenomenon from the perspective of modality bias and reveal that an excessive modality bias on the audio caused by dropout is the underlying reason. Moreover, we present the Modality Bias Hypothesis (MBH) to systematically describe the relationship between modality bias and robustness against missing modality in multimodal systems. Building on these findings, we propose a novel Multimodal Distribution Approximation with Knowledge Distillation (MDA-KD) framework to reduce over-reliance on the audio modality and to maintain performance and robustness simultaneously. Finally, to address an entirely missing modality, we adopt adapters to dynamically switch decision strategies. The effectiveness of our proposed approach is evaluated and validated through a series of comprehensive experiments using the MISP2021 and MISP2022 datasets. Our code is available at https://github.com/dalision/ModalBiasAVSR

cs.SD↗

arXiv2024

Adversarial speech for voice privacy protection from Personalized Speech generation

The rapid progress in personalized speech generation technology, including personalized text-to-speech (TTS) and voice conversion (VC), poses a challenge in distinguishing between generated and real speech for human listeners, resulting in an urgent demand in protecting speakers' voices from malicious misuse. In this regard, we propose a speaker protection method based on adversarial attacks. The proposed method perturbs speech signals by minimally altering the original speech while rendering downstream speech generation models unable to accurately generate the voice of the target speaker. For validation, we employ the open-source pre-trained YourTTS model for speech generation and protect the target speaker's speech in the white-box scenario. Automatic speaker verification (ASV) evaluations were carried out on the generated speech as the assessment of the voice protection capability. Our experimental results show that we successfully perturbed the speaker encoder of the YourTTS model using the gradient-based I-FGSM adversarial perturbation method. Furthermore, the adversarial perturbation is effective in preventing the YourTTS model from generating the speech of the target speaker. Audio samples can be found in https://voiceprivacy.github.io/Adeversarial-Speech-with-YourTTS.

eess.AS↗