Shilin Zhou

Zhengyu Liang, Yingqian Wang, Longguang Wang et al.

Exploiting spatial-angular correlation is crucial to light field (LF) image super-resolution (SR), but is highly challenging due to its non-local property caused by the disparities among LF images. Although many deep neural networks (DNNs) have been developed for LF image SR and achieved continuously improved performance, existing methods cannot well leverage the long-range spatial-angular correlation and thus suffer a significant performance drop when handling scenes with large disparity variations. In this paper, we propose a simple yet effective method to learn the non-local spatial-angular correlation for LF image SR. In our method, we adopt the epipolar plane image (EPI) representation to project the 4D spatial-angular correlation onto multiple 2D EPI planes, and then develop a Transformer network with repetitive self-attention operations to learn the spatial-angular correlation by modeling the dependencies between each pair of EPI pixels. Our method can fully incorporate the information from all angular views while achieving a global receptive field along the epipolar line. We conduct extensive experiments with insightful visualizations to validate the effectiveness of our method. Comparative results on five public datasets show that our method not only achieves state-of-the-art SR performance, but also performs robust to disparity variations. Code is publicly available at https://github.com/ZhengyuLiang24/EPIT.

Xinyi Ying, Li Liu, Yingqian Wang et al.

Training a convolutional neural network (CNN) to detect infrared small targets in a fully supervised manner has gained remarkable research interests in recent years, but is highly labor expensive since a large number of per-pixel annotations are required. To handle this problem, in this paper, we make the first attempt to achieve infrared small target detection with point-level supervision. Interestingly, during the training phase supervised by point labels, we discover that CNNs first learn to segment a cluster of pixels near the targets, and then gradually converge to predict groundtruth point labels. Motivated by this "mapping degeneration" phenomenon, we propose a label evolution framework named label evolution with single point supervision (LESPS) to progressively expand the point label by leveraging the intermediate predictions of CNNs. In this way, the network predictions can finally approximate the updated pseudo labels, and a pixel-level target mask can be obtained to train CNNs in an end-to-end manner. We conduct extensive experiments with insightful visualizations to validate the effectiveness of our method. Experimental results show that CNNs equipped with LESPS can well recover the target masks from corresponding point labels, {and can achieve over 70% and 95% of their fully supervised performance in terms of pixel-level intersection over union (IoU) and object-level probability of detection (Pd), respectively. Code is available at https://github.com/XinyiYing/LESPS.

Xinyi Ying, Li Liu, Zaipin Lin et al.

Multi-frame infrared small target (MIRST) detection in satellite videos is a long-standing, fundamental yet challenging task for decades, and the challenges can be summarized as: First, extremely small target size, highly complex clutters & noises, various satellite motions result in limited feature representation, high false alarms, and difficult motion analyses. Second, the lack of large-scale public available MIRST dataset in satellite videos greatly hinders the algorithm development. To address the aforementioned challenges, in this paper, we first build a large-scale dataset for MIRST detection in satellite videos (namely IRSatVideo-LEO), and then develop a recurrent feature refinement (RFR) framework as the baseline method. Specifically, IRSatVideo-LEO is a semi-simulated dataset with synthesized satellite motion, target appearance, trajectory and intensity, which can provide a standard toolbox for satellite video generation and a reliable evaluation platform to facilitate the algorithm development. For baseline method, RFR is proposed to be equipped with existing powerful CNN-based methods for long-term temporal dependency exploitation and integrated motion compensation & MIRST detection. Specifically, a pyramid deformable alignment (PDA) module and a temporal-spatial-frequency modulation (TSFM) module are proposed to achieve effective and efficient feature alignment, propagation, aggregation and refinement. Extensive experiments have been conducted to demonstrate the effectiveness and superiority of our scheme. The comparative results show that ResUNet equipped with RFR outperforms the state-of-the-art MIRST detection methods. Dataset and code are released at https://github.com/XinyiYing/RFR.

Lei Zhang, Zhenghua Li, Shilin Zhou et al.

Inspired by early research on exploring naturally annotated data for Chinese word segmentation (CWS), and also by recent research on integration of speech and text processing, this work for the first time proposes to mine word boundaries from parallel speech/text data. First we collect parallel speech/text data from two Internet sources that are related with CWS data used in our experiments. Then, we obtain character-level alignments and design simple heuristic rules for determining word boundaries according to pause duration between adjacent characters. Finally, we present an effective complete-then-train strategy that can better utilize extra naturally annotated data for model training. Experiments demonstrate our approach can significantly boost CWS performance in both cross-domain and low-resource scenarios.

Jiaqing Liu, Chong Deng, Qinglin Zhang et al.

Automatic Speech Recognition (ASR) transcripts exhibit recognition errors and various spoken language phenomena such as disfluencies, ungrammatical sentences, and incomplete sentences, hence suffering from poor readability. To improve readability, we propose a Contextualized Spoken-to-Written conversion (CoS2W) task to address ASR and grammar errors and also transfer the informal text into the formal style with content preserved, utilizing contexts and auxiliary information. This task naturally matches the in-context learning capabilities of Large Language Models (LLMs). To facilitate comprehensive comparisons of various LLMs, we construct a document-level Spoken-to-Written conversion of ASR Transcripts Benchmark (SWAB) dataset. Using SWAB, we study the impact of different granularity levels on the CoS2W performance, and propose methods to exploit contexts and auxiliary information to enhance the outputs. Experimental results reveal that LLMs have the potential to excel in the CoS2W task, particularly in grammaticality and formality, our methods achieve effective understanding of contexts and auxiliary information by LLMs. We further investigate the effectiveness of using LLMs as evaluators and find that LLM evaluators show strong correlations with human evaluations on rankings of faithfulness and formality, which validates the reliability of LLM evaluators for the CoS2W task.

Shilin Zhou, Zhenghua Li

Contextual Automatic Speech Recognition (ASR) faces challenges with large-scale keyword dictionaries, as excessive irrelevant candidates introduce noise that degrades accuracy. To address this, dynamic filtering typically uses a base ASR model to generate preliminary hypotheses, followed by semantic text retrievers to fetch a concise subset of relevant keywords. However, this approach frequently fails in Chinese ASR. Base models often produce homophonic or near-homophonic errors that preserve the phonetic cues of the target keywords but severely distort their semantic meaning, rendering standard semantic retrievers ineffective. To resolve this, we propose a filtering framework that jointly integrates Semantic, Pinyin, and Glyph features (JSPG). Pinyin effectively retrieves targets based on phonetic similarity, while glyph provides complementary structural cues to filter out numerous irrelevant homophones inherent in Chinese. To bridge the gap between character-level pinyin/glyph metrics and sequence-level filtering, we introduce an extended Smith-Waterman algorithm that computes similarity scores between the N-best hypothesis sequences and keywords. Experiments on the Aishell-1 and RWCS-NER datasets demonstrate that JSPG significantly outperforms single-feature baselines. Furthermore, downstream contextual ASR models guided by JSPG achieve substantial improvements in keyword recognition accuracy.

Z-Image Team, Huanqia Cai, Sihan Cao et al.

The landscape of high-performance image generation models is currently dominated by proprietary systems, such as Nano Banana Pro and Seedream 4.0. Leading open-source alternatives, including Qwen-Image, Hunyuan-Image-3.0 and FLUX.2, are characterized by massive parameter counts (20B to 80B), making them impractical for inference, and fine-tuning on consumer-grade hardware. To address this gap, we propose Z-Image, an efficient 6B-parameter foundation generative model built upon a Scalable Single-Stream Diffusion Transformer (S3-DiT) architecture that challenges the "scale-at-all-costs" paradigm. By systematically optimizing the entire model lifecycle -- from a curated data infrastructure to a streamlined training curriculum -- we complete the full training workflow in just 314K H800 GPU hours (approx. $630K). Our few-step distillation scheme with reward post-training further yields Z-Image-Turbo, offering both sub-second inference latency on an enterprise-grade H800 GPU and compatibility with consumer-grade hardware (<16GB VRAM). Additionally, our omni-pre-training paradigm also enables efficient training of Z-Image-Edit, an editing model with impressive instruction-following capabilities. Both qualitative and quantitative experiments demonstrate that our model achieves performance comparable to or surpassing that of leading competitors across various dimensions. Most notably, Z-Image exhibits exceptional capabilities in photorealistic image generation and bilingual text rendering, delivering results that rival top-tier commercial models, thereby demonstrating that state-of-the-art results are achievable with significantly reduced computational overhead. We publicly release our code, weights, and online demo to foster the development of accessible, budget-friendly, yet state-of-the-art generative models.

Xinyi Ying, Chao Xiao, Ruojing Li et al.

Small object detection (SOD) has been a longstanding yet challenging task for decades, with numerous datasets and algorithms being developed. However, they mainly focus on either visible or thermal modality, while visible-thermal (RGBT) bimodality is rarely explored. Although some RGBT datasets have been developed recently, the insufficient quantity, limited category, misaligned images and large target size cannot provide an impartial benchmark to evaluate multi-category visible-thermal small object detection (RGBT SOD) algorithms. In this paper, we build the first large-scale benchmark with high diversity for RGBT SOD (namely RGBT-Tiny), including 115 paired sequences, 93K frames and 1.2M manual annotations. RGBT-Tiny contains abundant targets (7 categories) and high-diversity scenes (8 types that cover different illumination and density variations). Note that, over 81% of targets are smaller than 16x16, and we provide paired bounding box annotations with tracking ID to offer an extremely challenging benchmark with wide-range applications, such as RGBT fusion, detection and tracking. In addition, we propose a scale adaptive fitness (SAFit) measure that exhibits high robustness on both small and large targets. The proposed SAFit can provide reasonable performance evaluation and promote detection performance. Based on the proposed RGBT-Tiny dataset and SAFit measure, extensive evaluations have been conducted, including 23 recent state-of-the-art algorithms that cover four different types (i.e., visible generic detection, visible SOD, thermal SOD and RGBT object detection). Project is available at https://github.com/XinyiYing/RGBT-Tiny.

Xinyi Ying, Yingqian Wang, Longguang Wang et al.

Infrared small target super-resolution (SR) aims to recover reliable and detailed high-resolution image with high-contrast targets from its low-resolution counterparts. Since the infrared small target lacks color and fine structure information, it is significant to exploit the supplementary information among sequence images to enhance the target. In this paper, we propose the first infrared small target SR method named local motion and contrast prior driven deep network (MoCoPnet) to integrate the domain knowledge of infrared small target into deep network, which can mitigate the intrinsic feature scarcity of infrared small targets. Specifically, motivated by the local motion prior in the spatio-temporal dimension, we propose a local spatio-temporal attention module to perform implicit frame alignment and incorporate the local spatio-temporal information to enhance the local features (especially for small targets). Motivated by the local contrast prior in the spatial dimension, we propose a central difference residual group to incorporate the central difference convolution into the feature extraction backbone, which can achieve center-oriented gradient-aware feature extraction to further improve the target contrast. Extensive experiments have demonstrated that our method can recover accurate spatial dependency and improve the target contrast. Comparative results show that MoCoPnet can outperform the state-of-the-art video SR and single image SR methods in terms of both SR performance and target enhancement. Based on the SR results, we further investigate the influence of SR on infrared small target detection and the experimental results demonstrate that MoCoPnet promotes the detection performance. The code is available at https://github.com/XinyiYing/MoCoPnet.

Feng Zhang, Xueying Wang, Shilin Zhou et al.

Rotated object detection in aerial images has received increasing attention for a wide range of applications. However, it is also a challenging task due to the huge variations of scale, rotation, aspect ratio, and densely arranged targets. Most existing methods heavily rely on a large number of pre-defined anchors with different scales, angles, and aspect ratios, and are optimized with a distance loss. Therefore, these methods are sensitive to anchor hyper-parameters and easily suffer from performance degradation caused by boundary discontinuity. To handle this problem, in this paper, we propose a dense anchor-free rotated object detector (DARDet) for rotated object detection in aerial images. Our DARDet directly predicts five parameters of rotated boxes at each foreground pixel of feature maps. We design a new alignment convolution module to extracts aligned features and introduce a PIoU loss for precise and stable regression. Our method achieves state-of-the-art performance on three commonly used aerial objects datasets (i.e., DOTA, HRSC2016, and UCAS-AOD) while keeping high efficiency. Code is available at https://github.com/zf020114/DARDet.

Zhengyu Liang, Yingqian Wang, Longguang Wang et al.

Light field (LF) image super-resolution (SR) aims at reconstructing high-resolution LF images from their low-resolution counterparts. Although CNN-based methods have achieved remarkable performance in LF image SR, these methods cannot fully model the non-local properties of the 4D LF data. In this paper, we propose a simple but effective Transformer-based method for LF image SR. In our method, an angular Transformer is designed to incorporate complementary information among different views, and a spatial Transformer is developed to capture both local and long-range dependencies within each sub-aperture image. With the proposed angular and spatial Transformers, the beneficial information in an LF can be fully exploited and the SR performance is boosted. We validate the effectiveness of our angular and spatial Transformers through extensive ablation studies, and compare our method to recent state-of-the-art methods on five public LF datasets. Our method achieves superior SR performance with a small model size and low computational cost. Code is available at https://github.com/ZhengyuLiang24/LFT.

Feng Zhang, Xueying Wang, Shilin Zhou et al.

Ship detection in remote sensing images plays a crucial role in various applications and has drawn increasing attention in recent years. However, existing arbitrary-oriented ship detection methods are generally developed on a set of predefined rotated anchor boxes. These predefined boxes not only lead to inaccurate angle predictions but also introduce extra hyper-parameters and high computational cost. Moreover, the prior knowledge of ship size has not been fully exploited by existing methods, which hinders the improvement of their detection accuracy. Aiming at solving the above issues, in this paper, we propose a center-head point extraction based detector (named CHPDet) to achieve arbitrary-oriented ship detection in remote sensing images. Our CHPDet formulates arbitrary-oriented ships as rotated boxes with head points which are used to determine the direction. And rotated Gaussian kernel is used to map the annotations into target heatmaps. Keypoint estimation is performed to find the center of ships. Then, the size and head point of the ships are regressed. The orientation-invariant model (OIM) is also used to produce orientation-invariant feature maps. Finally, we use the target size as prior to finetune the results. Moreover, we introduce a new dataset for multi-class arbitrary-oriented ship detection in remote sensing images at a fixed ground sample distance (GSD) which is named FGSD2021. Experimental results on FGSD2021 and two other widely used data sets, i.e., HRSC2016, and UCAS-AOD demonstrate that our CHPDet achieves state-of-the-art performance and can well distinguish between bow and stern. Code and FGSD2021 dataset are available at https://github.com/zf020114/CHPDet.

Che Liu, Yingji Zhang, Dong Zhang et al.

This work proposes an industry-level omni-modal large language model (LLM) pipeline that integrates auditory, visual, and linguistic modalities to overcome challenges such as limited tri-modal datasets, high computational costs, and complex feature alignments. Our pipeline consists of three main components: First, a modular framework enabling flexible configuration of various encoder-LLM-decoder architectures. Second, a lightweight training strategy that pre-trains audio-language alignment on the state-of-the-art vision-language model Qwen2.5-VL, thus avoiding the costly pre-training of vision-specific modalities. Third, an audio synthesis pipeline that generates high-quality audio-text data from diverse real-world scenarios, supporting applications such as Automatic Speech Recognition and Speech-to-Speech chat. To this end, we introduce an industry-level omni-modal LLM, Nexus. Extensive experiments validate the efficacy of our pipeline, yielding the following key findings:(1) In the visual understanding task, Nexus exhibits superior performance compared with its backbone model - Qwen2.5-VL-7B, validating the efficiency of our training strategy. (2) Within the English Spoken Question-Answering task, the model achieves better accuracy than the same-period competitor (i.e, MiniCPM-o2.6-7B) in the LLaMA Q. benchmark. (3) In our real-world ASR testset, Nexus achieves outstanding performance, indicating its robustness in real scenarios. (4) In the Speech-to-Text Translation task, our model outperforms Qwen2-Audio-Instruct-7B. (5) In the Text-to-Speech task, based on pretrained vocoder (e.g., Fishspeech1.4 or CosyVoice2.0), Nexus is comparable to its backbone vocoder on Seed-TTS benchmark. (6) An in-depth analysis of tri-modal alignment reveals that incorporating the audio modality enhances representational alignment between vision and language.

Shilin Zhou, Zhenghua Li

While end-to-end Automatic Speech Recognition (ASR) models have shown impressive performance in transcribing general speech, they often struggle to accurately recognize contextually relevant keywords, such as proper nouns or user-specific entities. Previous approaches have explored leveraging keyword dictionaries in the textual modality to improve keyword recognition, either through token-level fusion that guides token-by-token generation or phrase-level fusion that enables direct copying of keyword phrases. However, these methods operate at different granularities and have their own limitations. In this paper, we propose a novel multi-grained fusion approach that jointly leverages the strengths of both token-level and phrase-level fusion with Large Language Models (LLMs). Our approach incorporates a late-fusion strategy that elegantly combines ASR's acoustic information with LLM's rich contextual knowledge, balancing fine-grained token precision with holistic phrase-level understanding. Experiments on Chinese and English datasets demonstrate that our approach achieves state-of-the-art performance on keyword-related metrics while preserving high accuracy on non-keyword text. Ablation studies further confirm that the token-level and phrase-level components both contribute significantly to the performance gains, complementing each other in our joint multi-grained framework. The code and models will be publicly available at https://github.com/.

Zhichao Sheng, Shilin Zhou, Chen Gong et al.

Large Audio Language Models (LALMs), powered by the chain-of-thought (CoT) paradigm, have shown remarkable reasoning capabilities. Intuitively, different problems often require varying depths of reasoning. While some methods can determine whether to reason for a given problem, they typically lack a fine-grained mechanism to modulate how much to reason. This often results in a ``one-size-fits-all'' reasoning depth, which generates redundant overthinking for simple questions while failing to allocate sufficient thought to complex ones. In this paper, we conduct an in-depth analysis of LALMs and find that an effective and efficient LALM should reason smartly by adapting its reasoning depth to the problem's complexity. To achieve this, we propose a difficulty-adaptive reasoning method for LALMs. Specifically, we propose a reward function that dynamically links reasoning length to the model's perceived problem difficulty. This reward encourages shorter, concise reasoning for easy tasks and more elaborate, in-depth reasoning for complex ones. Extensive experiments demonstrate that our method is both effective and efficient, simultaneously improving task performance and significantly reducing the average reasoning length. Further analysis on reasoning structure paradigm offers valuable insights for future work.

Zhichao Sheng, Shilin Zhou, Chen Gong et al.

Spoken Language Understanding (SLU) plays a crucial role in speech-centric multimedia applications, enabling machines to comprehend spoken language in scenarios such as meetings, interviews, and customer service interactions. SLU encompasses multiple tasks, including Automatic Speech Recognition (ASR), spoken Named Entity Recognition (NER), and spoken Sentiment Analysis (SA). However, existing methods often rely on separate model architectures for individual tasks such as spoken NER and SA, which increases system complexity, limits cross-task interaction, and fails to fully exploit heterogeneous datasets available across tasks. To address these limitations, we propose UniSLU, a unified framework that jointly models multiple SLU tasks within a single architecture. Specifically, we propose a unified representation for diverse SLU tasks, enabling full utilization of heterogeneous datasets across multiple tasks. Built upon this representation, we propose a unified generative method that jointly models ASR, spoken NER, and SA tasks, enhancing task interactions and enabling seamless integration with large language models to harness their powerful generative capabilities. Extensive experiments on public SLU datasets demonstrate the effectiveness of our approach, achieving superior SLU performance compared to several benchmark methods, making it well-suited for real-world speech-based multimedia scenarios. We will release all code and models at github to facilitate future research.

Xuebin Wang, Lei Zhang, Zhenghua Li et al.

Inspired by early research on exploring naturally annotated data for Chinese Word Segmentation (CWS), and also by recent research on integration of speech and text processing, this work for the first time proposes to explicitly mine word boundaries from speech-text parallel data. We employ the Montreal Forced Aligner (MFA) toolkit to perform character-level alignment on speech-text data, giving pauses as candidate word boundaries. Based on detailed analysis of collected pauses, we propose an effective probability-based strategy for filtering unreliable word boundaries. To more effectively utilize word boundaries as extra training data, we also propose a robust complete-then-train (CTT) strategy. We conduct cross-domain CWS experiments on two target domains, i.e., ZX and AISHELL2. We have annotated about 1,000 sentences as the evaluation data of AISHELL2. Experiments demonstrate the effectiveness of our proposed approach.

Shilin Zhou, Zhenghua Li, Yu Hong et al.

End-to-end automatic speech recognition (ASR) systems have made significant progress in general scenarios. However, it remains challenging to transcribe contextual named entities (NEs) in the contextual ASR scenario. Previous approaches have attempted to address this by utilizing the NE dictionary. These approaches treat entities as individual tokens and generate them token-by-token, which may result in incomplete transcriptions of entities. In this paper, we treat entities as indivisible wholes and introduce the idea of copying into ASR. We design a systematic mechanism called CopyNE, which can copy entities from the NE dictionary. By copying all tokens of an entity at once, we can reduce errors during entity transcription, ensuring the completeness of the entity. Experiments demonstrate that CopyNE consistently improves the accuracy of transcribing entities compared to previous approaches. Even when based on the strong Whisper, CopyNE still achieves notable improvements.

Shilin Zhou, Qingrong Xia, Zhenghua Li et al.

This paper proposes to cast end-to-end span-based SRL as a word-based graph parsing task. The major challenge is how to represent spans at the word level. Borrowing ideas from research on Chinese word segmentation and named entity recognition, we propose and compare four different schemata of graph representation, i.e., BES, BE, BIES, and BII, among which we find that the BES schema performs the best. We further gain interesting insights through detailed analysis. Moreover, we propose a simple constrained Viterbi procedure to ensure the legality of the output graph according to the constraints of the SRL structure. We conduct experiments on two widely used benchmark datasets, i.e., CoNLL05 and CoNLL12. Results show that our word-based graph parsing approach achieves consistently better performance than previous results, under all settings of end-to-end and predicate-given, without and with pre-trained language models (PLMs). More importantly, our model can parse 669/252 sentences per second, without and with PLMs respectively.

Yu Zhang, Qingrong Xia, Shilin Zhou et al.

Semantic role labeling (SRL) is a fundamental yet challenging task in the NLP community. Recent works of SRL mainly fall into two lines: 1) BIO-based; 2) span-based. Despite ubiquity, they share some intrinsic drawbacks of not considering internal argument structures, potentially hindering the model's expressiveness. The key challenge is arguments are flat structures, and there are no determined subtree realizations for words inside arguments. To remedy this, in this paper, we propose to regard flat argument spans as latent subtrees, accordingly reducing SRL to a tree parsing task. In particular, we equip our formulation with a novel span-constrained TreeCRF to make tree structures span-aware and further extend it to the second-order case. We conduct extensive experiments on CoNLL05 and CoNLL12 benchmarks. Results reveal that our methods perform favorably better than all previous syntax-agnostic works, achieving new state-of-the-art under both end-to-end and w/ gold predicates settings.

Yi Hou, Hong Zhang, Shilin Zhou

Deep convolutional neural networks (CNN) have recently been shown in many computer vision and pattern recog- nition applications to outperform by a significant margin state- of-the-art solutions that use traditional hand-crafted features. However, this impressive performance is yet to be fully exploited in robotics. In this paper, we focus one specific problem that can benefit from the recent development of the CNN technology, i.e., we focus on using a pre-trained CNN model as a method of generating an image representation appropriate for visual loop closure detection in SLAM (simultaneous localization and mapping). We perform a comprehensive evaluation of the outputs at the intermediate layers of a CNN as image descriptors, in comparison with state-of-the-art image descriptors, in terms of their ability to match images for detecting loop closures. The main conclusions of our study include: (a) CNN-based image representations perform comparably to state-of-the-art hand- crafted competitors in environments without significant lighting change, (b) they outperform state-of-the-art competitors when lighting changes significantly, and (c) they are also significantly faster to extract than the state-of-the-art hand-crafted features even on a conventional CPU and are two orders of magnitude faster on an entry-level GPU.