Past ArXiv / August 18, 2026

Daily archive

A preserved paper digest from the previous month.

Archive date 08/18/2026

Source ArXiv

Back to current digest

Reading Queue

ArXiv

26 papers

Multimodal LLM

1 VibeWorlding: Can Multimodal Agents Construct 3D Open Worlds End-to-End? Yansong Ning, Jingwen Ye, Zhongkai Wu, Yang Sun, Yiqin Zhu, Xingyi Li, Weidong Zhang, Hao Liu

Multimodal LLM Benchmark & Evaluation Embodied Interaction LLM Interpretability

ArXiv

Paper 1 / arXiv:2608.15265 Open arXiv

Relevance 9 Novelty 7

Why selected: Matches criterion 3 very closely: an embodied/multimodal agent benchmark and RL training framework for constructing interactive 3D open worlds.

Constructing an interactive 3D open world from a user query is important. However, existing methods are primarily evaluated on idealized, simple queries, making it difficult to systematically analyze and compare how multimodal agents understand user intent, use 3D tools, and reason over textual and visual 3D world information. To this end, we propose VibeWorlding, a unified framework for benchmarking and training vibe worlding agents: a multimodal agent that can autonomously infer user intent, plan scene layout, invoke 3D tools, and reflect on the multimodal feedback in a multi-turn agent-environment interaction process. To achieve this, we first build VWE-BENCH, a benchmark of 2,616 high-quality 3D assets, 323 human-annotated seed 3D worlds, and 6,828 reverse-synthesized multimodal user queries, split into verified queries with ground-truth and unverified queries with carefully designed rubrics. Moreover, we develop VibeWorlding-Gym, a joint multimodal RL post-training framework that integrates (1) a sandbox environment unifying asset retrieval, editing, and image rendering as MCP tools, and (2) a rubric-based verifier that combines physical feasibility and intent fulfillment verification, supporting both fair model evaluation and scalable multimodal RL reward service. Our experiments show that current frontier MLLMs are far from solving the vibe worlding agent task, with even GPT-5.5 and Qwen3.8-Max reaching below 60% success rate, and trace the bottleneck to precise 3D world editing. We further find that RL training can ease this weakness and enable open-source MLLMs to even surpass closed-source frontiers: our VibeWorlder-8B is comparable to frontier MLLMs, while our flagship VibeWorlder-30B-A3B attains the best overall Pass@1 among all evaluated models.

7 PersonaDrive: Controllable Trajectory Prediction with Multi-Dimensional Driving Personas Chan Lee, Kimin Yun, Yuseok Bae, Seong Tae Kim, Jung Uk Kim

Multimodal LLM Benchmark & Evaluation Time-Series LLM

ArXiv

Paper 7 / arXiv:2608.15230 Open arXiv

Relevance 6 Novelty 5

Why selected: Matches criterion 1 closely: proposes controllable trajectory prediction with multi-dimensional driving personas for embodied driving behavior.

Although recent trajectory prediction and end-to-end autonomous driving methods improve robustness in urban environments, they still lack meaningful controllability. Existing benchmarks either provide no persona-conditioned annotations or support only a single urgency spectrum (i.e., emergency, normal, relaxed), which cannot distinguish personas that share the same urgency level but require different driving dynamics. To address this, we propose (i) the Persona-Conditioned Trajectory (PCT) dataset, which decomposes driving personas along two axes, Temporal Urgency and Ride Comfort, and combines three levels of each to form a grid of nine personas, each paired with natural-language descriptions and trajectories, and (ii) PersonaDrive, a framework that can learn driving personas from language and can generate persona-specific trajectories. PersonaDrive incorporates Persona-Conditioned Anchor Transform (PCAT), which hierarchically reshapes anchors along both axes, and Persona-Conditioned Multi-Modal Fusion (PCMF) for BEV-level persona fusion. Training is supervised by a Hierarchical Guide Loss enforcing axis-aligned physical orderings and an Axis-Decomposed Diversity Loss preventing diagonal mode collapse. Experimental results show that PersonaDrive consistently improves over the compared baselines across multi-dimensional scenarios. The code and PCT dataset are available at https://github.com/VisualAIKHU/PersonaDrive

10 Remote-Sensing City Layout Extraction with MLLM Zigan Zhou, Kai Li, Yupeng Deng

Multimodal LLM Generative Modeling Safety & Backdoors Retrieval & Grounding

ArXiv

Paper 10 / arXiv:2608.16484 Open arXiv

Relevance 5 Novelty 6

Why selected: Matches criterion 2 closely: it uses an MLLM for remote-sensing layout extraction and code generation, i.e. a new multimodal large language model application with structured visual reasoning.

Remote-sensing systems usually describe urban content with detection boxes, semantic masks, or vector boundaries. Such outputs locate classes and support image-plane scoring, yet they do not by themselves constitute an executable layout that retains object identities, typed relations, topology, and regeneration rules. Code-as-City instead casts urban-layout extraction from a single top-down image as constrained code generation with a multimodal large language model (MLLM). An image model first produces an aligned five-class semantic layout prior. Three ordered MLLM passes use the image and this prior to recover roads, land-cover regions and relations, and buildings. Deterministic normalization converts the accumulated records into a city graph and a restricted layout program. Executing the program creates a renderable 3D city layout and an orthographic semantic projection over shared geometry. The projection admits pixel-level comparison with remote-sensing masks, while named objects, relations, and editing operations remain available for synchronized regeneration of both views. Evaluated on the 100 scenes of CityLayout-100, the complete framework obtains 41.1% mean intersection-over-union and 48.3% global intersection-over-union. This result provides quantitative evidence that visual observations can be translated into inspectable, editable city code with coupled planar and 3D outputs.

13 TRACE-Bench: Decomposing and Diagnosing Multi-Reference Image Generation Haoran Wang, Chaofan Ma, Ran Yi, Lizhuang Ma

Multimodal LLM Benchmark & Evaluation Generative Modeling

ArXiv

Paper 13 / arXiv:2608.16765 Open arXiv

Relevance 5 Novelty 6

Why selected: Matches criterion 2 and 4: a benchmark for multi-reference image generation with a compositional diagnostic formulation, useful for evaluating multimodal generative models.

Despite recent advances in unified multimodal models for multi-reference image generation, existing benchmarks remain organized around predefined task types (e.g., "subject composition"), which are ill-suited to this combinatorial setting and lead to fragmented coverage, uncontrolled complexity, and little diagnostic value. Recognizing that diverse multi-reference tasks share a common set of atomic operations, we adopt a capability-oriented perspective and formalize four operators: Anchor ($f$), Disentangle ($g$), Apply ($\oplus$), and Compose ($C$). Any multi-reference prompt can then be represented as a compositional formula over these operators, whose structural complexity is quantified by the number of operator slots. Building on this formulation, we construct TRACE-Bench, comprising approximately 1,600 evaluation cases across slot counts 1--8, built from 631 formula templates and around 4,000 reference images spanning diverse artistic styles and real-world subjects. The formula structure directly drives an operator-aligned evaluation protocol for per-capability scoring and a diagnostic tree analysis for recursive failure localization. Evaluating 9 leading models reveals insights invisible to holistic scoring: the primary bottleneck lies in disentanglement ($g$) and attribute binding ($\oplus$) rather than scene-level composition ($C$), with even the best model scoring only 0.74 on attribute fidelity. Project page: https://amuseum-whr.github.io/TraceBench

Benchmark & Evaluation

2 Emergent 3D Instance Segmentation from Self-Supervised Point Transformers Ted Lentsch, Santiago Montiel-Mar\'in, Holger Caesar, Julian F. P. Kooij

Benchmark & Evaluation Safety & Backdoors

ArXiv

Paper 2 / arXiv:2608.15796 Open arXiv

Relevance 8 Novelty 7

Why selected: Strong match to criterion 1: it studies emergent spatial/3D instance structure in self-supervised point transformers and turns that into a prior-free 3D segmentation method.

Unsupervised 3D instance segmentation of outdoor LiDAR scans has traditionally relied on handcrafted geometric priors such as density-based clustering, motion cues, or projected 2D detections. In this work, we investigate whether a frozen, self-supervised point transformer already contains the structural information required to isolate object instances without any handcrafted geometric prior. Using this transformer purely as a feature extractor, we probe its internal representations across the SemanticKITTI, nuScenes, and Waymo Perception datasets. Our analysis yields four core insights: (1) the instance signal concentrates in the attention queries and keys rather than in the values or final output features; (2) output features semantically collapse, merging adjacent same-class objects that the queries and keys keep distinct; (3) this instance signal is bimodal in depth, strongest at the shallowest and deepest encoder stages; and (4) this signal is driven predominantly by the rotary position encoding (RoPE), whose removal collapses its advantage. We put these findings into our method TokenGraph3D, a training-free segmenter that groups points via connected components on a key-similarity graph, using neither density-based clustering nor proximity priors. Under identical prior-free conditions, we substantially outperform output-feature baselines, making the emergent 3D instance structure visible.

5 HarnessEval-W: Agentifying the Evaluation of Visual Worlds Weiliang Chen, Haowen Sun, Jun Gao, Jiawei Chi, Hanyang Wang, Qiyu Dai, Yihao Li, Hao Li, Jingnan Gao, Yi-Hsin Hung, Xingzhuo Guo, Shangchen Miao, Zhiyuan Shi, Xiang Li, Fengrui Tian, Weihua Du, Ziqi Huang, Shenyuan Gao, Siqiao Huang, Mingyu Liu, Yifei Li, Shizun Wang, Xi Wang, Tianqi Zhang, Xue Luo, Xiyin Ren, Jinshan Ren, Xiaoyang Shen, Xiaobo Hu, Zhiyang Dou, Mingyu Ding, Yichao Yan, Xinchao Wang, Yizhou Wang, Shilong Liu, Wenzhao Zheng, Yueqi Duan, Yuan Gong, Ziwei Liu, Ming-Yu Liu, Jialong Wu, Jiangran Lyu, Fangfu Liu

Benchmark & Evaluation Embodied Interaction Retrieval & Grounding

ArXiv

Paper 5 / arXiv:2608.16859 Open arXiv

Relevance 5 Novelty 7

Why selected: Matches criterion 3 closely: it is a benchmark/evaluation paper for world models, with a new agentified evaluation pipeline for visual worlds.

A benchmark should deliver more than a scalar score: what makes an evaluation trustworthy is the reasoning that justifies the score. This is especially critical for world models, where judging a rollout requires understanding whether physics, causality, and world state evolve correctly. Humans spot such violations naturally, yet no existing benchmark automates this capability: metrics are computed brute-force, leaving no reasoning chain that can be examined or verified. We introduce HarnessEval-W, an agentified evaluation pipeline that brings the harness paradigm from the LLM ecosystem to world model benchmarking. Rather than applying a fixed rubric, HarnessEval-W interprets the context of each evaluation case, decomposes the evaluation question into measurable subproblems, and spawns specialized sub-agents, each equipped with tailored context and diagnostic tools to reason over its own subproblem. The parent agent then validates the gathered evidence and summarizes it into the final verdict. This hierarchical workflow turns every evaluation into a transparent evidence tree whose complete reasoning chain justifies the result. We apply HarnessEval-W to 18 representative world models over 330 evaluation cases. Its judgments closely align with human preferences while providing verifiable, fine-grained diagnoses of every generated rollout. We open-source the full pipeline as a live benchmark and invite the broad community to contribute to grow new skills and evaluation cases as world models evolve.

9 CoM$^3$eT: A foundation model for medical image analysis through federated, multidimensional context integration J. Raphael Sch\"afer, Kai Geissler, Till Nicke, Chiara Tappermann, Karoline Heber, Eike Petersen, Habib Mergan, Lars Ole Schwen, Nick Weiss, Annika Gerken, Jan Hendrik Moltz, Tom Bisson, Isil Dogan O, Tim-Rasmus Kiehl, Norman Zerbe, Sefer Elezkurtaj, Robin S. Mayer, Nadine Flinner, Peter Wild, Isabel Dahm, Felix Peisen, Heinrich von Busch, Robert Grimm, Sebastian Arndt, Lisa Siegler, Matthias Stefan May, Antje Prasse, Natalia Artysh, Fabian Kiessling, Johannes Lotz

Benchmark & Evaluation Medical AI Generative Modeling

ArXiv

Paper 9 / arXiv:2608.16268 Open arXiv

Relevance 5 Novelty 6

Why selected: Matches criterion 4 closely: a medical vision foundation model with federated training and cross-domain image analysis applications.

Medical foundation models improve generalization when training AI models with limited labeled data, but remain confined to a single specialty, such as pathology or radiology, and to either sparse or dense outputs, such as classification or segmentation. Here, we present CoM$^3$eT (Co-representation Multidimensional Multitask Medical Transformer), a medical vision foundation model that unifies pathology and radiology, sparse and dense predictions, and two- and higher-dimensional inputs by modeling multidimensional context with attention. CoM$^3$eT outperformed other medical foundation models in an open competition spanning five tomographic, four whole-specimen, and three two-dimensional datasets, covering sparse and dense prediction tasks as well as report generation. When adapted across diverse clinical applications, training fewer than 2.5% of parameters achieved performance comparable to full fine-tuning, enabling research without access to high-performance GPU clusters. Applied to federated learning across hospitals, this approach achieved performance comparable to pooled-data training over internet connections and with consumer-grade hardware.

14 GenRouter: Unified Workflow Routing for Agentic Image Generation Harold Haodong Chen, Zhiyu Hou, Wen-Jie Shu, Weilin Ruan, Yingjie Xu, Litao Guo, Ying-Cong Chen

Benchmark & Evaluation Embodied Interaction Generative Modeling LLM Interpretability

ArXiv

Paper 14 / arXiv:2608.16721 Open arXiv

Relevance 5 Novelty 6

Why selected: Matches criterion 4: this is a vision-generation routing system built on agentic image-generation workflows, with a new framework for adaptive workflow selection.

The rapid evolution of text-to-image (T2I) generation models has effectively solved the foundational challenge of raw pixel synthesis, shifting the community's focus toward fulfilling increasingly intricate user requests. While recent agentic image generation workflows enhance static inference with advanced capabilities like external knowledge retrieval and iterative reasoning, they mostly operate in isolated silos with fixed ``one-size-fits-all" topologies. This inevitably leads to severe compute-mismatch, where simple queries are forced through computationally heavy pipelines. To bridge this gap, we present GenRouter, the first unified workflow routing framework for agentic image generation. We first formulate GenCanvas, standardizing diverse agentic pipelines into a universal set of foundational primitives and executable templates. Operating over this unified space, GenRouter adaptively routes heterogeneous prompts to their optimal workflows via (i) demand profiling, (ii) experience matching, and (iii) Pareto filtering. Extensive experiments across diverse benchmarks demonstrate that GenRouter achieves superior visual alignment while reducing execution costs by over 95% and latency by 65% compared to heavyweight static pipelines. Furthermore, the system continuously self-evolves via accumulated experience, enabling robust zero-shot generalization that boosts performance and halves computational overhead.

15 Unlocking the Potential of Image Editing via Concept Scaling and Dense Supervision Long Cui, Xiaoqian Liu, Qi Qin, Yi Xin, Tao Lin, Jianguo Li, Linfeng Zhang

Benchmark & Evaluation Generative Modeling

ArXiv

Paper 15 / arXiv:2608.16812 Open arXiv

Relevance 4 Novelty 6

Why selected: Matches criterion 4 partially: image editing with concept scaling and dense supervision, relevant to vision foundation model applications and generative modeling.

Existing image editing frameworks predominantly follow the training paradigm of text-to-image diffusion models. However, extending this paradigm to image editing highlights two inherent discrepancies, specifically, the insufficient attention to edit concept granularity and the training inefficiency caused by sparse supervision signals. To address these issues, we establish a comprehensive hierarchical taxonomy featuring over 1,000 fine-grained edit concepts and build ConceptEdit-12M, a massive dataset of 12 million high-quality editing pairs via an improved synthesis framework. This library-driven approach effectively rectifies the distribution collapse of generated data while ensuring high data fidelity. Furthermore, we propose a dense supervision training strategy that synthesizes multiple non-interfering concepts into single image pairs. By providing richer learning signals, this strategy significantly enhances both training efficiency and overall model performance. Training results validate our strategy, significantly outperforming prior works. Finally, we present ConceptEdit-Bench, a granular evaluation suite designed to diagnose model capabilities across a vast array of real-world scenarios.

16 MegaParts: Scaling Part-Aware 3D Object Generation to 300 Parts via Token-Efficient Autoregressive Modeling Manwen Liao, Xinyu Lian, Jian Mao, Kaixu Chen, Li Luo, Jinghao Yan, Wanshui Gan, Qiao Yu, Weitian Zhang, Chunhua Shen, Guang Chen, Bo Dai, Xudong Xu, Zhaoyang Lyu

Benchmark & Evaluation Generative Modeling LLM Interpretability

ArXiv

Paper 16 / arXiv:2608.14783 Open arXiv

Relevance 4 Novelty 6

Why selected: Matches criterion 4 partially: vision/3D generative modeling with scalable token-efficient autoregressive generation, though not a foundation model paper per se.

Part-aware 3D object generation is essential for graphics applications such as controllable modeling, editing, and articulation, where objects are represented as coherent assemblies of semantic parts. However, existing part-aware generation methods, do not scale well to highly complex objects. As the number of parts increases, generating detailed geometry becomes prohibitively expensive in token length and memory. We introduce MegaParts, a scalable autoregressive 3D generation framework to address this challenge by combining structured sequence modeling with a token-efficient vector-quantized shape tokenizer. Our tokenizer learns discrete latent representations for part-level geometry by minimizing token usage subject to high-fidelity reconstruction, enabling adaptive-length tokenization based on geometric complexity. On top of this compact representation, we train a large language model to generate object bounding boxes, part bounding boxes, and part shape tokens within a unified structured sequence. Combined with efficient long-context training strategy, our token-efficient formulation scales to objects with up to 300 parts and sequence lengths up to 256k tokens. This substantially extends the scale of part-aware 3D generation while preserving compositional structure and enabling fine-grained part-level control. Our method achieves higher mesh quality than baseline autoregressive and diffusion models, showing that compressed discrete part tokens improve not only scalability but also the achievable fidelity of generated geometry. These results suggest that LLM native token-efficient autoregressive modeling is a compelling alternative to diffusion for large-scale part-aware 3D generation. The project page is available at https://expmaster.github.io/megaparts_webpage.

22 CM-MAE: A Physics-Guided Cross-Modal Self-Supervised Learning Framework for Vision-Wireless Applications Yubo Zhang, Yiyao Liu

Benchmark & Evaluation LLM Interpretability

ArXiv

Paper 22 / arXiv:2608.15972 Open arXiv

Relevance 4 Novelty 5

Why selected: Matches criterion 4: a new self-supervised cross-modal pretraining method for vision-wireless representation learning, with a notable statistical trick via soft contrastive alignment.

Synchronized camera and wireless measurements observe the same scene through different physical channels. The central difficulty is that a representation learned in one deployment can fail when viewpoint, traffic, illumination, and propagation geometry change. This paper presents CM-MAE, a self-supervised vision--wireless pretraining framework for cross-scenario representation transfer. The evaluated real-data model uses only RGB frames and the measured 64-beam received-power vector available in DeepSense 6G; it does not use ray-traced paths, calibrated depth, or beam-index labels during pretraining. Its central pretraining term is a \emph{soft contrastive alignment loss}. Instead of making the synchronized image--wireless pair the only positive pair, this loss builds a target distribution from similarities between measured beam-power profiles, so nonidentical samples with similar directional responses are not forced apart as false negatives. A masked joint decoder provides the complementary local objective by reconstructing hidden visual patches and wireless angular clusters under modality dropout. After pretraining, a differential-rate fine-tuning rule lets a new fusion head adapt quickly while the encoders move slowly. Under a sequence-disjoint DeepSense 6G protocol, adding the soft alignment loss improves a matched linear-probe transfer average from 24.88\% to 29.49\%. Mild fusion fine-tuning reaches 77.38\% Top-1 accuracy on unseen Scenarios 6--8, and optional transductive normalization adaptation reaches 78.69\%. Since the fusion setting uses the contemporaneous 64-beam power vector at inference, these results should be read as representation-transfer diagnostics, not as proactive beam-prediction or reduced-sweeping claims.

24 BASeg: Boundary-Aware Remote Sensing Segmentation with Structural Penalties Yuexi Song, Kailai Sun, Zhuoyu Wang, Mingyi He, Paul Pu Liang, Shenhao Wang, Jinhua Zhao

Benchmark & Evaluation Generative Modeling LLM Interpretability

ArXiv

Paper 24 / arXiv:2608.15683 Open arXiv

Relevance 3 Novelty 5

Why selected: No very close criterion match; it is remote-sensing segmentation with a new loss and benchmark, adjacent to vision applications but not embodied AI or VLMs.

Semantic segmentation is a core computer vision task in the remote sensing field, accelerating advancements in ur- ban development, agriculture, ecology, water resources, and environmental monitoring. However, recent methods usually struggle to capture fine-grained object features and bound- ary details. Besides, current widely used datasets often lack city morphology diversity and segmentation on generative im- ages remains largely unexplored. To address these issues, we propose a Mahalanobis-Angle Boundary Loss (MABL) that explicitly enhances boundary and shape consistency. MABL jointly models structural importance and boundary orientation through Mahalanobis distance-based weighting and angle- aware penalty. It can be readily integrated into diverse seg- mentation architectures and consistently improves their accu- racy. Built upon MABL, we introduce BASeg, a boundary- aware remote sensing segmentation framework with Struc- tural Penalties. BASeg integrates a Global Visual State Space module (GSM) with a Cross-Feature Fusion module (CFM) to capture both long-range contextual dependencies and fine- grained local details. Additionally, we establish a global 10- city benchmark dataset (GCD-25k) to facilitate accurate build- ing and road segmentation. Extensive experiments on four remote-sensing benchmarks demonstrate that BASeg consis- tently outperforms existing methods, achieving up to a 2.8% improvement in mIoU while producing more accurate object boundary segmentation across diverse scenes. Moreover, integrating MABL into multiple existing segmentation archi- tectures consistently improves performance across datasets, demonstrating its robustness and broad applicability.

Vision-Language Reasoning

3 MOSS-VL Technical Report Pengyu Wang, Chenkun Tan, Shaojun Zhou, Qirui Zhou, Yanxin Chen, Xingyang He, Huazheng Zeng, Jijun Cheng, Chenghao Wang, Xiaomeng Qian, Pengfei Wang, Zhan Huang, Shanqing Gao, Wei Huang, Longjun Cao, Wu Ran, Jie Liu, Changtai Zhu, Hongkai Wang, Yixian Tian, Chenghao Liu, Zhen Ye, Xinghao Wang, Botian Jiang, Guoguo Feng, Zhaoye Fei, Ruixiao Li, Mingshu Chen, Yang Gao, Qinyuan Cheng, Shimin Li, Xipeng Qiu

Vision-Language Reasoning Benchmark & Evaluation LLM Interpretability Time-Series LLM

ArXiv

Paper 3 / arXiv:2608.15045 Open arXiv

Relevance 8 Novelty 6

Why selected: Matches criterion 2 closely: a new open vision-language model family for real-time streaming interaction and proactive perception-while-speaking.

We present MOSS-VL, an open vision-language model family that treats real-time interaction -- perceiving while it speaks -- as a first-class capability. It is co-designed across the stack: the language decoder attends to vision only through gated cross-attention, so the model can naturally see incoming frames while generating; a synthesized interaction corpus supervises when to speak, when to stay silent, and when to revise; and a staged curriculum concentrates all real-time-specific training in one light final stage over a strong offline foundation. Offline, MOSS-VL-Instruct is competitive at comparable scale and leads temporal-reasoning video sets. Across four streaming benchmarks, MOSS-VL-Realtime posts the best average on three (second on the fourth) among open-source streaming models, sweeping the three subsets that squarely test proactive behavior -- 66.0 vs. 37.5 for the best baseline on OmniMMI Proactive Alerting. With 11.3B parameters but visual tokens outside the decoded sequence, MOSS-VL widens its time-to-first-token advantage over same-backbone Qwen3-VL-8B from 2.8x to 5.1x as visual context grows. We release all five checkpoints, the training curriculum, and the real-time inference code at https://github.com/OpenMOSS/MOSS-VL.

4 Beyond Similarity Matching: Structured Reasoning for Open-Vocabulary Referring Segmentation in 3DGS Yizhao Wang, Xinfa Wang, Jingbo Wang, Jingbo Wang, Guantao Zhang, Yafeng Han, Guohong Gao, Yuhe Xia

Vision-Language Reasoning Benchmark & Evaluation LLM Interpretability Retrieval & Grounding

ArXiv

Paper 4 / arXiv:2608.16103 Open arXiv

Relevance 6 Novelty 7

Why selected: Matches criterion 1 and 4: language-guided open-vocabulary referring segmentation in 3DGS with explicit spatial/part-relation reasoning.

Open-vocabulary referring segmentation in 3D Gaussian Splatting (3DGS) requires a neural model to select Gaussian primitives according to free-form language expressions. Existing 3DGS-based methods usually rely on global text-region similarity, which is weak for queries involving attributes, reference objects, spatial relations, and fine-grained parts. This often causes target-reference confusion, granularity mismatch, part-whole leakage, and relation violations. We propose QAGaussian, a query-adaptive neural reasoning framework for language-guided Gaussian primitive selection. QAGaussian first learns query-conditioned multi-scale Gaussian slots as differentiable candidates whose receptive fields are shaped by the input expression. It then builds a relation-aware slot graph with language-conditioned edge weighting to propagate target-reference, attribute, part-whole, and contextual evidence. A granularity-adaptive router softly combines region-level, object-level, part-level, attribute-aware, and relation-aware mask branches, followed by relation-constrained refinement for spatial, part-whole, attribute, and geometric consistency. QAGaussian is pretrained only on Mosaic3D-5.6M for Gaussian-text alignment and evaluated on independent benchmarks without target-dataset fine-tuning. It achieves 47.2 Avg. mIoU and 63.2 Avg. F1, outperforming the strongest 3DGS referring baseline by 2.7 mIoU points and 2.9 F1 points. It also improves Part-mIoU from 38.6 to 43.4, Rel-mIoU from 44.4 to 50.8, and reduces target-reference confusion from 10.8 to 7.4. These results demonstrate that query-conditioned slot learning, relation-aware graph reasoning, and adaptive routing provide an effective neural modeling strategy for open-vocabulary referring segmentation in 3DGS. The code is available at https://github.com/zqeslwyz/QAGaussian.

8 MetaReason: Precise Interleaved Multimodal Reasoning via Editing Meta Information for Solving Geometry Problems Penghao Yin, Haomin Wang, Qihong Tang, Xiaoye Qu, Hongjie Zhang, Xiao-Ping Zhang

Vision-Language Reasoning Multimodal LLM Benchmark & Evaluation

ArXiv

Paper 8 / arXiv:2608.15006 Open arXiv

Relevance 6 Novelty 5

Why selected: Matches criterion 2 closely: introduces a new multimodal VLM-style reasoning framework with interleaved visual/text reasoning and a new benchmark.

Although visual reasoning is crucial for solving complex geometry tasks, existing vision-language models rely heavily on text-only reasoning. Some recent methods introduce intermediate visual states to facilitate reasoning, but they are often hindered by inaccurate geometric representations and low rendering fidelity, ultimately leading to unreliable outputs. To address these limitations, we propose MetaReason, a framework for multimodal reasoning in plane geometry that leverages structured meta-information to enable accurate auxiliary-line construction. The framework first parses geometric images into meta-information, performs controllable edits with predefined tools to synthesize high-fidelity visual states, and then conducts reasoning based on these augmented views. To support this framework, we construct TutorGeo, a comprehensive dataset containing 17k image-to-meta conversion samples, 60k text-only reasoning traces, and 60k interleaved multimodal reasoning traces. Using this dataset, we combine supervised fine-tuning and reinforcement learning to develop robust multimodal reasoning capabilities. We also introduce ExamGeo, a benchmark derived from real-world examination problems that enables systematic evaluation across varying difficulty levels. Experimental results demonstrate that MetaReason significantly outperforms existing open-source models and achieves competitive performance against proprietary models.

12 PixelControl: Fine-Grained Condition Fidelity in Text-to-Image Diffusion Xin Lin, Haodong Li, Zhifei Zhang, Yutong Yang, Haitian Zheng, Juanxi Tian, Zhe Lin, Truong Nguyen

Vision-Language Reasoning Generative Modeling LLM Interpretability

ArXiv

Paper 12 / arXiv:2608.15705 Open arXiv

Relevance 5 Novelty 6

Why selected: Matches criterion 4: a controllable text-to-image diffusion method focused on fine-grained spatial condition fidelity and boundary preservation.

Controllable text-to-image diffusion models can often follow the global layout of spatial conditions, yet still violate fine-grained structures such as object boundaries, thin contours, and medium/small conditioned regions. This limitation is especially problematic for VAE-based latent diffusion, where spatial compression can weaken high-frequency and low-area condition signals. We propose PixelControl, a pixel-space controllable diffusion framework for fine-grained condition fidelity. Built on a PixelDiT-style backbone, PixelControl avoids the latent bottleneck and introduces two complementary designs. First, Structure-Aware Control Injection derives a condition structure map and uses it to strengthen injected control residuals around spatially sensitive regions. Second, Multi-Scale Pyramid Cycle Loss verifies generated images against condition-derived structures across multiple resolutions, balancing global layout consistency with local boundary and detail accuracy. PixelControl supports depth, segmentation, edge, and their combinations through modality-specific control branches with lightweight gated fusion. Experiments across depth, segmentation, and edge control show that PixelControl improves structural fidelity and visual quality over existing controllable generation methods, with especially strong gains on boundaries and medium/small conditioned regions. The project page can be found at: https://linxin0.github.io/pixelcontrol_homepage/pixelcontrol-site/

17 Beyond Accuracy: Assessing Calibration of Geospatial Foundation Models and Their Sensitivity to Distribution Shifts Nils Lehmann, Jakob Gawlikowski, Burak Ekim, Isaac Corley, Xiao Xiang Zhu

Vision-Language Reasoning Benchmark & Evaluation

ArXiv

Paper 17 / arXiv:2608.16614 Open arXiv

Relevance 5 Novelty 5

Why selected: Matches criterion 4: a careful evaluation of geospatial foundation models, emphasizing calibration and distribution shift rather than just accuracy.

Geospatial Foundation Models (GeoFMs) are most commonly ranked and selected by accuracy on standard benchmark conditions via averaged ranks. We show that this protocol is too narrow: the promised deployment in critical EO tasks requires further angles of analysis, mainly calibration, the agreement between a model's confidence and its correctness. Across 16 frozen encoders, four classification and five segmentation datasets, and two orthogonal stress axes, every encoder degrades as corruption intensifies, and the ranking changes as well. Across the four classification benchmarks, EO-pretrained and ImageNet-pretrained encoders are indistinguishable on clean accuracy and clean calibration, and EO pretraining provides no more stability under shift than ImageNet pretraining. Under shift the GeoFMs drift further into overconfidence than the ImageNet-pretrained encoders, at every grade and in every corruption family. A centered kernel alignment (CKA) analysis ties this to representational rigidity: EO-pretrained embeddings move less under corruption while losing just as much task information and remaining overconfident. We apply three commonly explored uncertainty quantification methods and find that temperature scaling and deep ensembles cannot counteract the degradation, while a Gaussian-process probe roughly halves ECE under severe cloud only by tripling it on clean data. In selective prediction experiments, we find that confidence-based abstention cannot defer around confidently wrong predictions, and advocate that benchmark rankings and evaluations should therefore operate across a multitude of conditions and metrics to more holistically evaluate model development progress and close the gap to real world deployment scenarios.

21 ConceptFormer: Learning Adaptive Latent Concepts for Query-Document Alignment in Visual Document Retrieval Peng Chunyi, Xu Zhipeng, Yan Yukun, Liu Zhenghao, Yu Shi, Mei Sen, Sun Yubo, Zhang Yongheng, Zhou Jie, Gu Yu, Yu Ge, Sun Maosong

Vision-Language Reasoning Multimodal LLM Benchmark & Evaluation Generative Modeling

ArXiv

Paper 21 / arXiv:2608.15698 Open arXiv

Relevance 4 Novelty 5

Why selected: Matches criterion 4 closely: vision foundation model methods applied to visual document retrieval, with latent concept alignment.

Visual document retrieval is a critical component of multimodal retrieval-augmented generation, aiming to identify query-relevant pages from document collections where evidence is distributed across text, layout, charts, and visual structures. Recent efforts toward finer-grained supervision primarily rely on textual descriptions or localized visual regions as evidence proxies. However, such supervision signals may either overlook complex visual structures or provide incomplete and inaccurate representations of the underlying evidence. To address these limitations, we propose ConceptFormer, a latent concept representation learning framework for visual document retrieval. ConceptFormer models query-relevant evidence as continuous, query-conditioned latent concepts that explicitly bridge localized visual evidence and semantic relevance, without requiring either textual intermediate representations or direct reliance on raw visual annotations. During training, ConceptFormer employs a strong vision-language model to dynamically determine the number of latent concept tokens and uses these concepts as an intermediate representation to bridge the semantic gap between queries and documents, thereby guiding the learning of the embedding space. Experiments on diverse visual document retrieval benchmarks demonstrate that ConceptFormer achieves 16.7\% and 22.1\% relative improvements in average NDCG@10 over the strongest visual retrieval baseline and the strongest OCR-based text retrieval baseline, respectively. Further analysis reveals that latent concepts effectively connect localized visual evidence with semantic relevance, enabling the retriever to capture both fine-grained textual cues and complex document-level visual structures while preserving strong retrieval alignment. Codes and data are available at https://github.com/Neuir/ConceptFormer.

23 CLARA: Clip-Level Multimodal Alignment with VLM-Derived Rationales for Hateful Video Detection Yuchen Zhang, Shuang Dai, Zeyu Fu, Yunfei Long, Ravi Shekhar, Haralambos Mouratidis

Vision-Language Reasoning Multimodal LLM Benchmark & Evaluation Safety & Backdoors

ArXiv

Paper 23 / arXiv:2608.15905 Open arXiv

Relevance 3 Novelty 5

Why selected: Matches criterion 2 only loosely through VLM-derived rationales, but the main task is hateful video detection rather than a new VLLM/MLLM.

Hateful video detection has become increasingly important with the rapid growth of video-centric social media platforms, given the serious risks that hate speech poses to both individual well-being and social cohesion. Compared with text or static multimodal content, hateful video detection remains underexplored and significantly more challenging, as hateful meaning often arises from complex interactions among multimodal cues, including speech, audio, and visual content. Moreover, such signals are often brief, implicit, and temporally dependent, making them difficult to capture using conventional video-level representations. In this work, we propose CLARA, a clip-level multimodal framework for hateful video detection. Instead of treating a video as a single instance, CLARA models it as a sequence of fine-grained clips, enabling more precise capture of temporally localized hateful signals. We introduce a Mixture-of-Experts clip encoder for adaptive multimodal alignment, a local-global segment contrastive objective to jointly model short-term cues and long-range temporal dependencies, and VLM-derived rationales integrated via a gated Transformer to provide high-level semantic guidance. Extensive experiments on three hateful video datasets demonstrate that CLARA consistently outperforms state-of-the-art methods. Further ablation studies and parameter analyses validate the effectiveness of each component.

25 Depth-Dominant Skeleton Detection for Natural Scenes Chengkun Rao, Yixuan Deng, Min Li, Yangjun Ou, Ye Li, Ziwei Luo, Zhaojing Wang, Junwei Tang, Bangchao Wang, Xiaoyun Yan

Vision-Language Reasoning Benchmark & Evaluation Safety & Backdoors

ArXiv

Paper 25 / arXiv:2608.16367 Open arXiv

Relevance 3 Novelty 5

Why selected: No direct criterion match, though it is a spatial perception method; the depth-dominant idea is about skeleton detection in natural scenes rather than embodied spatial intelligence.

To date, all natural scene skeleton detection follows the paradigm of taking RGB images as the sole input; despite notable progress, methods under this paradigm suffer significant performance degradation on complex-content images. We observe that depth images are inherently insensitive to color and texture, and can provide clear regional contours and inter-region spatial relationships, which naturally alleviates the difficulty of skeleton detection in complex scenarios. Motivated by this observation, this paper proposes for the first time a novel skeleton detection paradigm where depth images serve as the dominant modality and RGB images act as the auxiliary, and accordingly presents a model DDSkel (short for Depth-Dominant Skeleton Detection) under this paradigm. DDSkel employs an asymmetric encoder design to fuse RGB information into depth features, with the RGB modality branch having only 12% the parameters of the depth modality branch. DDSkel has a simple structure without intricate designs. Nevertheless, with only 36% of the trainable parameters of the current best method, DDSkel outperforms all state-of-the-art approaches on SymPASCAL, the most challenging dataset with a large volume of complex images.

26 Hierarchical Adaptive Feature Refinement Network for VHR Remote Sensing Image Segmentation Shuaishuai Cao, Meng Tang, Shuwei Peng, Xuan Liu, Min Huang, Jie Chen, Jiacheng Niu, Yong Chen, Edore Akpokodje, Hui Lin

Vision-Language Reasoning LLM Interpretability

ArXiv

Paper 26 / arXiv:2608.15647 Open arXiv

Relevance 3 Novelty 4

Why selected: No very close criterion match; it is a remote-sensing segmentation method with pretrained hierarchical encoders, which is adjacent to criterion 4 but not really about vision foundation models themselves.

Semantic segmentation of very-high-resolution (VHR) remote sensing imagery increasingly benefits from strong pretrained hierarchical encoders, yet exploiting their multi-stage representations remains difficult. Nearby regions demand different balances between fine detail and semantic context, aggressive task-specific transformations perturb useful pretrained features, and conventional semantic supervision provides limited structural guidance. We present HAFR-Net, a progressive refinement framework that adaptively organizes and conservatively refines hierarchical representations instead of replacing them with a monolithic decoder transformation. Heterogeneity-Guided Stage-Adaptive Fusion (HG-SAF) predicts dense stage weights conditioned on local feature variation. A Frequency-Residual Adapter (FRA) then injects frequency information through a bounded, zero-initialized residual branch that keeps the fused representation as its reference. A Confusion-Aware Tri-Prior Decoder (CATP) finally regularizes the prediction with boundary, objectness, and training-derived class-relation cues. Under a matched Swin-B training and single-scale inference protocol, HAFR-Net attains 84.12%, 87.86%, 55.17%, and 67.70% mIoU on ISPRS Vaihingen, ISPRS Potsdam, LoveDA, and OpenEarthMap, improving the matched UPerNet baseline by 0.55, 0.95, 1.55, and 1.84 percentage points, respectively. Controlled analyses further show consistent spatial reweighting beyond content-only routing, improved boundary and thin-structure accuracy over matched spatial and spectral alternatives, and reduced confusion on pre-declared class pairs.

Generative Modeling

6 SplatGuide: Geometric Priors from 3D Gaussians for Pose-Free Novel View Synthesis Yejun Zhang, Zihan Wang, Xu Ji, Yihao Wang, Yuxin Hou, Junyuan Fang, Juho-Matti Kilpel\"ainen, Arno Solin, Hamed Rezazadegan Tavakoli, Esa Rahtu, Juho Kannala

Generative Modeling

ArXiv

Paper 6 / arXiv:2608.16863 Open arXiv

Relevance 6 Novelty 6

Why selected: Matches criterion 4: pose-free novel view synthesis using 3D Gaussian priors and multi-view diffusion, which is directly in vision foundation model / 3D reconstruction applications.

Generating photorealistic novel views from unposed images requires both 3D geometric understanding and the ability to synthesize unseen content. A natural strategy combines feed-forward 3DGS reconstruction with multi-view diffusion. Yet prior pipelines extract at most one signal from the reconstruction, either pixel rendering or learned features, while none exploits per-Gaussian visibility for occlusion-aware reference selection. This *information disconnect* leaves renderable geometry, visibility cues, and learned features unused. SplatGuide closes this disconnect by reusing a single 3DGS scene across three complementary roles. Rendered images provide pixel-aligned geometric conditioning. Per-Gaussian source-view indices are rendered into a target-view voting map for occlusion-aware reference selection. Reconstruction tokens supply feature-level guidance via cross-attention. All three signals derive from the same reconstruction forward pass. Across RealEstate10K, DL3DV, Tanks-and-Temples, and Mip-NeRF 360, SplatGuide achieves state-of-the-art pose-free novel view synthesis. On RealEstate10K, with a moderate number of input views, it surpasses the ground-truth-pose baseline.

11 An Empirical Study of Training Pixel-Space Text-to-Image Diffusion Models Dengyang Jiang, Ruoyi Du, Zhennan Chen, Dongyang Liu, Zanyi Wang, Mingzhe Zheng, Xiangpeng Yang, Huanqia Cai, Aiming Hao, Yuming Jiang, Peng Gao, Harry Yang, Steven Hoi

Generative Modeling LLM Interpretability

ArXiv

Paper 11 / arXiv:2608.16887 Open arXiv

Relevance 4 Novelty 7

Why selected: No direct criterion match; it is a generative modeling paper, but focused on pixel-space text-to-image diffusion training rather than a new multimodal model or embodied AI.

This paper investigates an increasingly important topic in generative modeling: pixel-space diffusion models. Although numerous studies have explored this topic, most focus on small-scale or class-conditional settings. Consequently, a practical recipe for training pixel-space models that rival or exceed well-established latent-space counterparts remains elusive. Through a comprehensive empirical study, we first observe that direct large-scale pre-training in pixel space converges substantially more slowly than in latent space. This observation motivates a latent-to-pixel strategy that acquires generative priors efficiently in latent space and transitions to pixel space during post-training. We then systematically investigate the key design choices governing this transition, including weight initialization, data composition, prediction target, decoder architecture, and noise schedule, and identify a practical recipe that makes the resulting pixel-space models match or outperform their latent-space counterparts while delivering 3.18 to 4.75 times end-to-end inference speedups. We hope that our findings provide useful empirical insights and practical guidelines for future research on pixel-space generation.

19 Scalable Black-Box Model Attribution for Images Asaf Livne, Amir Jevnisek, Shai Avidan

Generative Modeling

ArXiv

Paper 19 / arXiv:2608.15652 Open arXiv

Relevance 3 Novelty 6

Why selected: No very close match to the listed criteria; it is more about black-box image model attribution than embodied AI, VLLMs/MLLMs, or vision foundation model applications.

The rapid proliferation of generative models raises the model attribution problem: given only an image, can we determine which model produced it? Existing methods have grown as elaborate as the generators they target, on the as- sumption that a more sophisticated model demands a more sophisticated attributor. We show it does not. RPA (Raw- Patch Attribution) attributes images in the strictest black- box setting with a lightweight CNN. Despite its simplicity, it attributes more models at higher accuracy than prior work, reaching 98.0% on 25-class DRAGON and 92.9% on 27- class OpenFake; it is data-efficient and runs at a cost inde- pendent of the number of candidate models; and it stays ro- bust to the compression, blur, and resizing images undergo in the wild. Training for closed-set attribution yields a ver- satile feature extractor: the same representation recovers model lineage without supervision, flags and groups unseen generators, and admits new models through few-shot adap- tation rather than retraining.

20 ES3D: Embedding Semantics into 3D Space for Component-Aware Editing Xuancheng Jin, Rengan Xie, Jiayuan Lu, Wenting Zheng, Rui Wang, Yuchi Huo, Lincheng Li, Yingfeng Chen

Generative Modeling Retrieval & Grounding

ArXiv

Paper 20 / arXiv:2608.15749 Open arXiv

Relevance 4 Novelty 5

Why selected: Matches criterion 4 partially: uses multimodal semantics for component-aware 3D editing, relevant to vision foundation model applications.

Existing 3D editing methods have made notable progress in controllability, yet they remain limited in several important ways. Most approaches rely on text-driven editing, which struggles to express fine-grained visual changes intended by the user. Moreover, many methods require manually supplied 3D masks or introduce unintended changes to regions that should remain untouched. These limitations largely arise from the absence of fine-grained semantic understanding, making it difficult for existing models to retrieve or modify specific 3D components. We introduce ES3D, a framework that embeds semantics directly into 3D space, enabling component-aware retrieval and editing of a 3D asset conditioned on multiple local reference images and optional text queries. We first construct a 3D semantic embedding by projecting multi-view semantic features into the voxelized space of the asset. We then perform 3D component retrieval by computing feature similarity between the 3D semantic embedding and the semantic embeddings of image or text queries. For editing, we employ a pretrained 3D generative model with an inpainting mechanism to modify the retrieved components guided by user-provided images while preserving the rest of the asset. Overall, ES3D is a 3D editing framework that retrieves editable regions based on semantic cues and uses multiple images as conditions. Extensive experiments demonstrate that ES3D produces geometrically consistent and semantically coherent edits, enabling robust image-based and text-assisted control for 3D editing.

General ML

18 Trajectory-Level Automatic Curriculum Learning for Legged Locomotion on Unstructured Terrain Rocky Liu, Tengyu Liu, Baoxiong Jia, Fangwei Zhong, Xinyi Tong, Hongzhao Xie, Siyuan Huang

General ML

ArXiv

Paper 18 / arXiv:2608.16164 Open arXiv

Relevance 5 Novelty 5

Why selected: Matches criterion 3: this is an embodied locomotion method with a novel trajectory-level automatic curriculum for unstructured terrain.

Training locomotion policies for complex unstructured terrain requires a curriculum to avoid early exploration failures. However, since unstructured terrain lacks explicit difficulty ordering for curriculum design, existing methods resort to heuristic curricula over parameterized terrains. This abstraction limits generalization, as policies can overadapt to near-fixed perceptual patterns. To address this, we propose \textbf{\ourname{}}, an \textbf{T}rajectory-level \textbf{A}utomatic \textbf{C}urriculum \textbf{L}earning framework that generates training tasks directly from unstructured terrain maps. At each curriculum update, the evaluator learns a difficulty function for the current policy that maps a given trajectory task to a difficulty score. The sampler then proposes new trajectories guided by the learned evaluator as the curriculum for the next policy update. This forms a closed loop in which the curriculum is iteratively matched to the evolving policy. Quantitative and qualitative experiments show that \ourname{} continuously provides effective curricula on unstructured terrain, improving trajectory success rate by $56.3\%$ over direct training without curriculum. Compared with handcrafted curriculum learning, our method improves success rate by $18.5\%$ on the hardest terrain tasks and by up to $39.74\%$ when evaluating traversal from diverse approach directions on the same obstacle type.

Archived papers 26

Top score 16

Average score 10.7