arXiv subjects

Zhen Li

Publications and source records attributed to Zhen Li.

At least 163 records · Page 9Linked to original sources

arXiv2025

SQS: Enhancing Sparse Perception Models via Query-based Splatting in Autonomous Driving

Sparse Perception Models (SPMs) adopt a query-driven paradigm that forgoes explicit dense BEV or volumetric construction, enabling highly efficient computation and accelerated inference. In this paper, we introduce SQS, a novel query-based splatting pre-training specifically designed to advance SPMs in autonomous driving. SQS introduces a plug-in module that predicts 3D Gaussian representations from sparse queries during pre-training, leveraging self-supervised splatting to learn fine-grained contextual features through the reconstruction of multi-view images and depth maps. During fine-tuning, the pre-trained Gaussian queries are seamlessly integrated into downstream networks via query interaction mechanisms that explicitly connect pre-trained queries with task-specific queries, effectively accommodating the diverse requirements of occupancy prediction and 3D object detection. Extensive experiments on autonomous driving benchmarks demonstrate that SQS delivers considerable performance gains across multiple query-based 3D perception tasks, notably in occupancy prediction and 3D object detection, outperforming prior state-of-the-art pre-training approaches by a significant margin (i.e., +1.3 mIoU on occupancy prediction and +1.0 NDS on 3D detection).

cs.CV↗

arXiv2025

IMD: A 6-DoF Pose Estimation Benchmark for Industrial Metallic Objects

Object 6DoF (6D) pose estimation is essential for robotic perception, especially in industrial settings. It enables robots to interact with the environment and manipulate objects. However, existing benchmarks on object 6D pose estimation primarily use everyday objects with rich textures and low-reflectivity, limiting model generalization to industrial scenarios where objects are often metallic, texture-less, and highly reflective. To address this gap, we propose a novel dataset and benchmark namely \textit{Industrial Metallic Dataset (IMD)}, tailored for industrial applications. Our dataset comprises 45 true-to-scale industrial components, captured with an RGB-D camera under natural indoor lighting and varied object arrangements to replicate real-world conditions. The benchmark supports three tasks, including video object segmentation, 6D pose tracking, and one-shot 6D pose estimation. We evaluate existing state-of-the-art models, including XMem and SAM2 for segmentation, and BundleTrack and BundleSDF for pose estimation, to assess model performance in industrial contexts. Evaluation results show that our industrial dataset is more challenging than existing household object datasets. This benchmark provides the baseline for developing and comparing segmentation and pose estimation algorithms that better generalize to industrial robotics scenarios.

cs.CV↗

arXiv2025

Augmented Reality-Enhanced Robot Teleoperation for Collecting User Demonstrations

Traditional industrial robot programming is often complex and time-consuming, typically requiring weeks or even months of effort from expert programmers. Although Programming by Demonstration (PbD) offers a more accessible alternative, intuitive interfaces for robot control and demonstration collection remain challenging. To address this, we propose an Augmented Reality (AR)-enhanced robot teleoperation system that integrates AR-based control with spatial point cloud rendering, enabling intuitive, contact-free demonstrations. This approach allows operators to control robots remotely without entering the workspace or using conventional tools like the teach pendant. The proposed system is generally applicable and has been demonstrated on ABB robot platforms, specifically validated with the IRB 1200 industrial robot and the GoFa 5 collaborative robot. A user study evaluates the impact of real-time environmental perception, specifically with and without point cloud rendering, on task completion accuracy, efficiency, and user confidence. Results indicate that enhanced perception significantly improves task performance by 28% and enhances user experience, as reflected by a 12% increase in the System Usability Scale (SUS) score. This work contributes to the advancement of intuitive robot teleoperation, AR interface design, environmental perception, and teleoperation safety mechanisms in industrial settings for demonstration collection. The collected demonstrations may serve as valuable training data for machine learning applications.

cs.RO↗

arXiv2025

Deep learning for 3D point cloud processing -- from approaches, tasks to its implications on urban and environmental applications

Point cloud processing as a fundamental task in the field of geomatics and computer vision, has been supporting tasks and applications at different scales from air to ground, including mapping, environmental monitoring, urban/tree structure modeling, automated driving, robotics, disaster responses etc. Due to the rapid development of deep learning, point cloud processing algorithms have nowadays been almost explicitly dominated by learning-based approaches, most of which are yet transitioned into real-world practices. Existing surveys primarily focus on the ever-updating network architecture to accommodate unordered point clouds, largely ignoring their practical values in typical point cloud processing applications, in which extra-large volume of data, diverse scene contents, varying point density, data modality need to be considered. In this paper, we provide a meta review on deep learning approaches and datasets that cover a selection of critical tasks of point cloud processing in use such as scene completion, registration, semantic segmentation, and modeling. By reviewing a broad range of urban and environmental applications these tasks can support, we identify gaps to be closed as these methods transformed into applications and draw concluding remarks in both the algorithmic and practical aspects of the surveyed methods.

cs.CV↗

arXiv2025

MiniCPM4: Ultra-Efficient LLMs on End Devices

This paper introduces MiniCPM4, a highly efficient large language model (LLM) designed explicitly for end-side devices. We achieve this efficiency through systematic innovation in four key dimensions: model architecture, training data, training algorithms, and inference systems. Specifically, in terms of model architecture, we propose InfLLM v2, a trainable sparse attention mechanism that accelerates both prefilling and decoding phases for long-context processing. Regarding training data, we propose UltraClean, an efficient and accurate pre-training data filtering and generation strategy, and UltraChat v2, a comprehensive supervised fine-tuning dataset. These datasets enable satisfactory model performance to be achieved using just 8 trillion training tokens. Regarding training algorithms, we propose ModelTunnel v2 for efficient pre-training strategy search, and improve existing post-training methods by introducing chunk-wise rollout for load-balanced reinforcement learning and data-efficient tenary LLM, BitCPM. Regarding inference systems, we propose CPM.cu that integrates sparse attention, model quantization, and speculative sampling to achieve efficient prefilling and decoding. To meet diverse on-device requirements, MiniCPM4 is available in two versions, with 0.5B and 8B parameters, respectively. Furthermore, we construct a hybrid reasoning model, MiniCPM4.1, which can be used in both deep reasoning mode and non-reasoning mode. Evaluation results demonstrate that MiniCPM4 and MiniCPM4.1 outperform similar-sized open-source models across benchmarks, with the 8B variants showing significant speed improvements on long sequence understanding and generation.

cs.CL↗

arXiv2025

The Kinematical Behavior of Solar Eruptive Filaments Affected by the Poloidal Magnetic Field

Kinematics of solar eruptive filaments is one of the important diagnostic parameters for predicting whether solar eruptions would induce geomagnetic storms. Particularly, some geomagnetic storms might be induced by solar filament eruptions originating from unexpected surface source regions because of non-radial ejection. The non-radial ejection of filaments has received widespread attention but remains inconclusive. We select two eruptive filaments, both of which are supported by flux ropes, as indicated by the hot channel structures seen in the 94 Å images and the hook-shaped brightenings where the filament material falls back. We measure the three-dimensional ejection trajectory of the eruptive filaments by integrating the simultaneous observations from SDO and STEREO. Furthermore, we calculate the distribution of the poloidal field along the ejection path and compare it to the ejection acceleration. It is revealed that the reinforcement of the poloidal magnetic field may lead to the suppression of the acceleration, with the acceleration resuming its increase only when the poloidal field diminishes to a certain level. Additionally, we compute the spatial distribution of the poloidal field in various directions and find that the poloidal magnetic field above the filaments is asymmetric. For both investigated events, the filaments appear to eject towards the side where the poloidal magnetic field is weaker, indicating that the eruptive filaments tend to propagate along the side with weaker strapping force. This may provide a new explanation for the inclined ejection of filaments.

astro-ph.SR↗

arXiv2025

OneVAE: Joint Discrete and Continuous Optimization Helps Discrete Video VAE Train Better

Encoding videos into discrete tokens could align with text tokens to facilitate concise and unified multi-modal LLMs, yet introducing significant spatiotemporal compression compared to continuous video representation. Previous discrete video VAEs experienced unstable training, long training time, and degraded reconstruction quality. Given the easier training and superior performance of continuous VAEs, an intuitive idea is to enhance discrete video VAEs by leveraging continuous VAEs. After rethinking the intrinsic link between discrete and continuous representations, we found that FSQ could effectively preserve pre-trained continuous VAE priors compared to other quantization methods. By leveraging continuous VAE priors, it converges several times faster than training from scratch and achieves superior performance at convergence. Meanwhile, two structural improvements are proposed. First, inspired by how continuous VAEs enhance reconstruction via enlarged latent dimensions, we introduce a multi-token quantization mechanism, which achieves nearly a 1 dB improvement in PSNR without compromising the token compression ratio. Second, to tackle reconstruction challenges in high-compression video VAEs, we strengthen first-frame reconstruction, enabling the causal VAE to leverage this information in subsequent frames and markedly improving the performance of 4 x 16 x 16 discrete VAEs. Furthermore, we propose a joint discrete-continuous optimization scheme that unifies the two paradigms and, for the first time, achieves competitive performance on both continuous and discrete representations within a single network. We name our method OneVAE to reflect this connection.

cs.CV↗

arXiv2025

AR-1-to-3: Single Image to Consistent 3D Object Generation via Next-View Prediction

Novel view synthesis (NVS) is a cornerstone for image-to-3d creation. However, existing works still struggle to maintain consistency between the generated views and the input views, especially when there is a significant camera pose difference, leading to poor-quality 3D geometries and textures. We attribute this issue to their treatment of all target views with equal priority according to our empirical observation that the target views closer to the input views exhibit higher fidelity. With this inspiration, we propose AR-1-to-3, a novel next-view prediction paradigm based on diffusion models that first generates views close to the input views, which are then utilized as contextual information to progressively synthesize farther views. To encode the generated view subsequences as local and global conditions for the next-view prediction, we accordingly develop a stacked local feature encoding strategy (Stacked-LE) and an LSTM-based global feature encoding strategy (LSTM-GE). Extensive experiments demonstrate that our method significantly improves the consistency between the generated views and the input views, producing high-fidelity 3D assets.

cs.CV↗

arXiv2025

MDK12-Bench: A Comprehensive Evaluation of Multimodal Large Language Models on Multidisciplinary Exams

Multimodal large language models (MLLMs), which integrate language and visual cues for problem-solving, are crucial for advancing artificial general intelligence (AGI). However, current benchmarks for measuring the intelligence of MLLMs suffer from limited scale, narrow coverage, and unstructured knowledge, offering only static and undifferentiated evaluations. To bridge this gap, we introduce MDK12-Bench, a large-scale multidisciplinary benchmark built from real-world K-12 exams spanning six disciplines with 141K instances and 6,225 knowledge points organized in a six-layer taxonomy. Covering five question formats with difficulty and year annotations, it enables comprehensive evaluation to capture the extent to which MLLMs perform over four dimensions: 1) difficulty levels, 2) temporal (cross-year) shifts, 3) contextual shifts, and 4) knowledge-driven reasoning. We propose a novel dynamic evaluation framework that introduces unfamiliar visual, textual, and question form shifts to challenge model generalization while improving benchmark objectivity and longevity by mitigating data contamination. We further evaluate knowledge-point reference-augmented generation (KP-RAG) to examine the role of knowledge in problem-solving. Key findings reveal limitations in current MLLMs in multiple aspects and provide guidance for enhancing model robustness, interpretability, and AI-assisted education.

cs.AI↗

arXiv2025

Time Evidence Fusion Network: Multi-source View in Long-Term Time Series Forecasting

In practical scenarios, time series forecasting necessitates not only accuracy but also efficiency. Consequently, the exploration of model architectures remains a perennially trending topic in research. To address these challenges, we propose a novel backbone architecture named Time Evidence Fusion Network (TEFN) from the perspective of information fusion. Specifically, we introduce the Basic Probability Assignment (BPA) Module based on evidence theory to capture the uncertainty of multivariate time series data from both channel and time dimensions. Additionally, we develop a novel multi-source information fusion method to effectively integrate the two distinct dimensions from BPA output, leading to improved forecasting accuracy. Lastly, we conduct extensive experiments to demonstrate that TEFN achieves performance comparable to state-of-the-art methods while maintaining significantly lower complexity and reduced training time. Also, our experiments show that TEFN exhibits high robustness, with minimal error fluctuations during hyperparameter selection. Furthermore, due to the fact that BPA is derived from fuzzy theory, TEFN offers a high degree of interpretability. Therefore, the proposed TEFN balances accuracy, efficiency, stability, and interpretability, making it a desirable solution for time series forecasting.

cs.LG↗

arXiv2025

LaTCoder: Converting Webpage Design to Code with Layout-as-Thought

Converting webpage designs into code (design-to-code) plays a vital role in User Interface (UI) development for front-end developers, bridging the gap between visual design and functional implementation. While recent Multimodal Large Language Models (MLLMs) have shown significant potential in design-to-code tasks, they often fail to accurately preserve the layout during code generation. To this end, we draw inspiration from the Chain-of-Thought (CoT) reasoning in human cognition and propose LaTCoder, a novel approach that enhances layout preservation in webpage design during code generation with Layout-as-Thought (LaT). Specifically, we first introduce a simple yet efficient algorithm to divide the webpage design into image blocks. Next, we prompt MLLMs using a CoTbased approach to generate code for each block. Finally, we apply two assembly strategies-absolute positioning and an MLLM-based method-followed by dynamic selection to determine the optimal output. We evaluate the effectiveness of LaTCoder using multiple backbone MLLMs (i.e., DeepSeek-VL2, Gemini, and GPT-4o) on both a public benchmark and a newly introduced, more challenging benchmark (CC-HARD) that features complex layouts. The experimental results on automatic metrics demonstrate significant improvements. Specifically, TreeBLEU scores increased by 66.67% and MAE decreased by 38% when using DeepSeek-VL2, compared to direct prompting. Moreover, the human preference evaluation results indicate that annotators favor the webpages generated by LaTCoder in over 60% of cases, providing strong evidence of the effectiveness of our method.

cs.SE↗

arXiv2025

Oedipus and the Sphinx: Benchmarking and Improving Visual Language Models for Complex Graphic Reasoning

Evaluating the performance of visual language models (VLMs) in graphic reasoning tasks has become an important research topic. However, VLMs still show obvious deficiencies in simulating human-level graphic reasoning capabilities, especially in complex graphic reasoning and abstract problem solving, which are less studied and existing studies only focus on simple graphics. To evaluate the performance of VLMs in complex graphic reasoning, we propose ReasonBench, the first evaluation benchmark focused on structured graphic reasoning tasks, which includes 1,613 questions from real-world intelligence tests. ReasonBench covers reasoning dimensions related to location, attribute, quantity, and multi-element tasks, providing a comprehensive evaluation of the performance of VLMs in spatial, relational, and abstract reasoning capabilities. We benchmark 11 mainstream VLMs (including closed-source and open-source models) and reveal significant limitations of current models. Based on these findings, we propose a dual optimization strategy: Diagrammatic Reasoning Chain (DiaCoT) enhances the interpretability of reasoning by decomposing layers, and ReasonTune enhances the task adaptability of model reasoning through training, all of which improves VLM performance by 33.5\%. All experimental data and code are in the repository: https://huggingface.co/datasets/cistine/ReasonBench.

cs.AI↗

arXiv2025

T2VParser: Adaptive Decomposition Tokens for Partial Alignment in Text to Video Retrieval

Text-to-video retrieval essentially aims to train models to align visual content with textual descriptions accurately. Due to the impressive general multimodal knowledge demonstrated by image-text pretrained models such as CLIP, existing work has primarily focused on extending CLIP knowledge for video-text tasks. However, videos typically contain richer information than images. In current video-text datasets, textual descriptions can only reflect a portion of the video content, leading to partial misalignment in video-text matching. Therefore, directly aligning text representations with video representations can result in incorrect supervision, ignoring the inequivalence of information. In this work, we propose T2VParser to extract multiview semantic representations from text and video, achieving adaptive semantic alignment rather than aligning the entire representation. To extract corresponding representations from different modalities, we introduce Adaptive Decomposition Tokens, which consist of a set of learnable tokens shared across modalities. The goal of T2VParser is to emphasize precise alignment between text and video while retaining the knowledge of pretrained models. Experimental results demonstrate that T2VParser achieves accurate partial alignment through effective cross-modal content decomposition. The code is available at https://github.com/Lilidamowang/T2VParser.

cs.CV↗

arXiv2025

LLMs are Also Effective Embedding Models: An In-depth Overview

Large language models (LLMs) have revolutionized natural language processing by achieving state-of-the-art performance across various tasks. Recently, their effectiveness as embedding models has gained attention, marking a paradigm shift from traditional encoder-only models like ELMo and BERT to decoder-only, large-scale LLMs such as GPT, LLaMA, and Mistral. This survey provides an in-depth overview of this transition, beginning with foundational techniques before the LLM era, followed by LLM-based embedding models through two main strategies to derive embeddings from LLMs. 1) Direct prompting: We mainly discuss the prompt designs and the underlying rationale for deriving competitive embeddings. 2) Data-centric tuning: We cover extensive aspects that affect tuning an embedding model, including model architecture, training objectives, data constructions, etc. Upon the above, we also cover advanced methods for producing embeddings from longer texts, multilingual, code, cross-modal data, as well as reasoning-aware and other domain-specific scenarios. Furthermore, we discuss factors affecting choices of embedding models, such as performance/efficiency comparisons, dense vs sparse embeddings, pooling strategies, and scaling law. Lastly, the survey highlights the limitations and challenges in adapting LLMs for embeddings, including cross-task embedding quality, trade-offs between efficiency and accuracy, low-resource, long-context, data bias, robustness, etc. This survey serves as a valuable resource for researchers and practitioners by synthesizing current advancements, highlighting key challenges, and offering a comprehensive framework for future work aimed at enhancing the effectiveness and efficiency of LLMs as embedding models.

cs.CL↗

arXiv2025

Yume: An Interactive World Generation Model

Yume aims to use images, text, or videos to create an interactive, realistic, and dynamic world, which allows exploration and control using peripheral devices or neural signals. In this report, we present a preview version of \method, which creates a dynamic world from an input image and allows exploration of the world using keyboard actions. To achieve this high-fidelity and interactive video world generation, we introduce a well-designed framework, which consists of four main components, including camera motion quantization, video generation architecture, advanced sampler, and model acceleration. First, we quantize camera motions for stable training and user-friendly interaction using keyboard inputs. Then, we introduce the Masked Video Diffusion Transformer~(MVDT) with a memory module for infinite video generation in an autoregressive manner. After that, training-free Anti-Artifact Mechanism (AAM) and Time Travel Sampling based on Stochastic Differential Equations (TTS-SDE) are introduced to the sampler for better visual quality and more precise control. Moreover, we investigate model acceleration by synergistic optimization of adversarial distillation and caching mechanisms. We use the high-quality world exploration dataset \sekai to train \method, and it achieves remarkable results in diverse scenes and applications. All data, codebase, and model weights are available on https://github.com/stdstu12/YUME. Yume will update monthly to achieve its original goal. Project page: https://stdstu12.github.io/YUME-Project/.

cs.CV↗

arXiv2025

Lumina-mGPT 2.0: Stand-Alone AutoRegressive Image Modeling

We present Lumina-mGPT 2.0, a stand-alone, decoder-only autoregressive model that revisits and revitalizes the autoregressive paradigm for high-quality image generation and beyond. Unlike existing approaches that rely on pretrained components or hybrid architectures, Lumina-mGPT 2.0 is trained entirely from scratch, enabling unrestricted architectural design and licensing freedom. It achieves generation quality on par with state-of-the-art diffusion models such as DALL-E 3 and SANA, while preserving the inherent flexibility and compositionality of autoregressive modeling. Our unified tokenization scheme allows the model to seamlessly handle a wide spectrum of tasks-including subject-driven generation, image editing, controllable synthesis, and dense prediction-within a single generative framework. To further boost usability, we incorporate efficient decoding strategies like inference-time scaling and speculative Jacobi sampling to improve quality and speed, respectively. Extensive evaluations on standard text-to-image benchmarks (e.g., GenEval, DPG) demonstrate that Lumina-mGPT 2.0 not only matches but in some cases surpasses diffusion-based models. Moreover, we confirm its multi-task capabilities on the Graph200K benchmark, with the native Lumina-mGPT 2.0 performing exceptionally well. These results position Lumina-mGPT 2.0 as a strong, flexible foundation model for unified multimodal generation. We have released our training details, code, and models at https://github.com/Alpha-VLLM/Lumina-mGPT-2.0.

cs.CV↗

arXiv2025

BleedOrigin: Dynamic Bleeding Source Localization in Endoscopic Submucosal Dissection via Dual-Stage Detection and Tracking

Intraoperative bleeding during Endoscopic Submucosal Dissection (ESD) poses significant risks, demanding precise, real-time localization and continuous monitoring of the bleeding source for effective hemostatic intervention. In particular, endoscopists have to repeatedly flush to clear blood, allowing only milliseconds to identify bleeding sources, an inefficient process that prolongs operations and elevates patient risks. However, current Artificial Intelligence (AI) methods primarily focus on bleeding region segmentation, overlooking the critical need for accurate bleeding source detection and temporal tracking in the challenging ESD environment, which is marked by frequent visual obstructions and dynamic scene changes. This gap is widened by the lack of specialized datasets, hindering the development of robust AI-assisted guidance systems. To address these challenges, we introduce BleedOrigin-Bench, the first comprehensive ESD bleeding source dataset, featuring 1,771 expert-annotated bleeding sources across 106,222 frames from 44 procedures, supplemented with 39,755 pseudo-labeled frames. This benchmark covers 8 anatomical sites and 6 challenging clinical scenarios. We also present BleedOrigin-Net, a novel dual-stage detection-tracking framework for the bleeding source localization in ESD procedures, addressing the complete workflow from bleeding onset detection to continuous spatial tracking. We compare with widely-used object detection models (YOLOv11/v12), multimodal large language models, and point tracking methods. Extensive evaluation demonstrates state-of-the-art performance, achieving 96.85% frame-level accuracy ($\pm\leq8$ frames) for bleeding onset detection, 70.24% pixel-level accuracy ($\leq100$ px) for initial source detection, and 96.11% pixel-level accuracy ($\leq100$ px) for point tracking.

cs.CV↗

arXiv2025

KptLLM++: Towards Generic Keypoint Comprehension with Large Language Model

The emergence of Multimodal Large Language Models (MLLMs) has revolutionized image understanding by bridging textual and visual modalities. However, these models often struggle with capturing fine-grained semantic information, such as the precise identification and analysis of object keypoints. Keypoints, as structure-aware, pixel-level, and compact representations of objects, particularly articulated ones, play a crucial role in applications such as fine-grained image analysis, object retrieval, and behavior recognition. In this paper, we propose KptLLM++, a novel multimodal large language model that specifically designed for generic keypoint comprehension through the integration of diverse input modalities guided by user-defined instructions. By unifying keypoint detection across varied contexts, KptLLM++ establishes itself as an advanced interface, fostering more effective human-AI collaboration. The model is built upon a novel identify-then-detect paradigm, which first interprets keypoint semantics and subsequently localizes their precise positions through a structured chain-of-thought reasoning mechanism. To push the boundaries of performance, we have scaled up the training dataset to over 500K samples, encompassing diverse objects, keypoint categories, image styles, and scenarios with complex occlusions. This extensive scaling enables KptLLM++ to unlock its potential, achieving remarkable accuracy and generalization. Comprehensive experiments on multiple keypoint detection benchmarks demonstrate its state-of-the-art performance, underscoring its potential as a unified solution for fine-grained image understanding and its transformative implications for human-AI interaction.

cs.CV↗