ScaleHP: Estimating Hand Pose in Metric Space

Accurate metric-space hand pose estimation (HPE) is essential for immersive human-computer interaction and robotics. However, most existing methods predict poses in a root-relative coordinate system and cannot estimate the hand in absolute metric scale. In this work, we observe that the intrinsic proportional relationships among human hand bones encode stable anthropometric priors that implicitly correlate with the overall metric size of the hand. Leveraging this insight, we present ScaleHP, an end-to-end one-stage hand pose estimation framework that bypasses fragile extrinsic depth modules to recover the hand in metric space. ScaleHP employs a transformer-based decoder with a novel scale token to fuse multi-scale morphological and appearance features. By solving for metric coordinates through a perspective-constrained least-squares approach, we achieve high-precision pose estimation in the camera coordinate system. ScaleHP delivers state-of-the-art performance, including 35.8 CS-MPJPE on FreiHand and 4.6/5.9 PA-MPJPE on DexYCB and HO3Dv3. These results demonstrate that internal biological constraints significantly reduce relative geometry and absolute metric errors, offering a robust solution for generalized, real-world hand tracking.

Dual Distribution Estimation for Zero-shot Noisy Test-Time Adaptation with VLMs

While test-time adaptation (TTA) empowers vision-language models to adapt without costly retraining, it remains highly vulnerable to out-of-distribution (OOD) outliers prevalent in real-world applications. This discrepancy motivates Noisy TTA (NTTA), an online task to filter noisy OOD samples on the fly while maximizing in-distribution (ID) classification accuracy. Existing zero-shot NTTA approaches typically rely on test-time discriminative training, leading to overconfident misclassifications and significantly degraded inference efficiency. To address these limitations, we propose a novel framework named Dual Distribution Estimation (DDE), shifting the zero-shot NTTA paradigm from instance-level learning to training-free Gaussian distribution modeling. DDE incorporates two novel modules: Positive Feature Distribution Estimation (PFDE) and Negative Label Distribution Estimation (NLDE). PFDE explicitly models class-wise inclusion and exclusion Gaussian distributions to formulate a calibrated contrastive score, robustly enhancing ID accuracy. In parallel, NLDE improves OOD identification by explicitly modeling the negative label distribution to mine highly discriminative labels, effectively mitigating spurious correlations. Extensive experiments show that on the large-scale ImageNet benchmark, DDE achieves an improvement of 3.70\% in harmonic mean accuracy and reduces the FPR95 for OOD detection by 6.20\%, while ensuring highly scalable and efficient online inference. Furthermore, DDE is zero-shot and training-free, demonstrating remarkable robustness in data-scarce scenarios. Codes are available at https://github.com/ZhuWenjie98/DDE.

Reflective VLA: In-Context Action Consequences Make VLAs Generalize

Most vision-language-action (VLA) models are reactive: they predict the next action from the current instruction and observation, implicitly assuming that the current observation fully specifies the action-relevant state. In embodied control, however, embodiment-specific factors such as camera-to-robot geometry, robot calibration, or systematic actuation bias are often hard to identify from a single observation. As a result, reactive policies cannot reliably disambiguate these factors in general, overfitting to training environments and generalizing poorly at deployment. We propose Reflective VLA, which conditions each decision on a context of observation-action-consequence triplets. Each triplet records not only what the robot observed and executed, but also how the scene changed afterward, exposing the deployment-specific mapping from actions to observed effects. Architecturally, Reflective VLA routes all observation modalities through the VLM under shared attention, so the action expert reasons directly over past triplets and the current observation. A block-causal mask enables parallel multi-frame training without leakage and supports KV-cached real-time inference. On standard LIBERO and SimplerEnv-Bridge, Reflective VLA preserves strong in-distribution performance. Under distribution shift on LIBERO-Plus and the harder LIBERO-Plus-Hard, it improves average success rate by 5.4 and 4.2 percentage points over a matched reactive baseline. Ablations with a matched history-only baseline further show that action consequences -- rather than additional context length alone -- are the key to cross-environment generalization. Project page: https://lianqing11.github.io/reflective-vla-page/

LemonHarness Technical Report

As large language model (LLM) agents are applied to longer tasks, they increasingly modify workspace state across multiple rounds of iteration. However, agents typically observe only tool outputs and log fragments, while the actual state changes occur in the file system. Without explicit workspace boundaries, state-changing operations such as file writes and temporary artifact generation may scatter changes across paths. Over time, these weakly constrained changes accumulate, making states such as modified files difficult to track. This paper presents LemonHarness, an integrated execution framework for long-horizon agents. LemonHarness establishes an explicit execution boundary by constraining state-changing operations within a clearly defined workspace and bringing model invocation, tool execution, and rule knowledge within a single controlled boundary. State-changing operations, including file writes, dependency installation, and temporary artifact creation, are executed through structured tool interfaces, with execution feedback recorded as observations available to subsequent model decisions. The system also introduces a reusable rule knowledge base, which turns recurring execution rules and acceptance criteria into runtime knowledge. LemonHarness further adds a time-aware execution mechanism that exposes elapsed and remaining budget to the model, so it can rebalance exploration, implementation, and validation effort as time pressure shifts and avoid timeouts from long waits or excessive verification. On Terminal-Bench 2.0, LemonHarness_GPT-5.3-CodeX reached 84.49% accuracy over 445 trials; pairing the same framework with the stronger GPT-5.5 backbone raised the average accuracy to 86.52% across five jobs. The results suggest that a unified runtime boundary, callable rule knowledge, and time-aware execution can improve the stability of long-horizon agent execution.

Tool-IQA: Augmenting Image Quality Assessment with Simple Tools

Vision-Language Models (VLMs) have been increasingly adopted for Image Quality Assessment (IQA). However, current methods typically employ a static one-shot scoring paradigm, despite the fact that humans assess image quality through dynamic visual inspection, e.g., selectively adjusting views to verify details and subtle artifacts. Specifically, relying solely on a single-pass observation introduces two primary limitations: first, perceiving the image only at a global scale restricts the assessment of finer local details; second, the original intensity distribution of the image may overwhelm the visibility, leading to insufficient inspection of image quality. To address these issues, we propose Tool-IQA, shifting the assessment mechanism from passive scoring to a tool-augmented workflow. In particular, we equip VLMs with simple yet effective view tools: a Magnifier to inspect local details, and a Gamma Corrector to uncover visibility and hidden artifacts. The assessment follows a structured pipeline that consists of an initial observation with rubric notes, a tool-augmented in-depth inspection, and a final quantification for calibrated quality score. Furthermore, to ensure efficient and purposeful tool callings, we introduce a batch-aware training strategy to reward tool interactions that can yield positive contributions rather than simply encouraging usage. Experiments on a variety of IQA benchmarks demonstrate that, with effective tool calling and calibrated assessment, our proposed Tool-IQA significantly outperforms existing state-of-the-art models, e.g., it achieves a PLCC of 0.854 on the challenging CLIVE dataset.

Text-Vision Co-Instructed Image Editing

Existing image editing methods can be generally categorized into textual instruction-based and visual prompt-based ones. Textual instructions are semantically expressive, but are limited by the coarse granularity of spatial control of the editing results. In contrast, visual prompts such as drag and point can provide precise spatial guidance, but are limited by the inherent ambiguity in semantic intent. To unify the strength of textual and visual prompts, we present Text-Vision Co-Instructed Image Editing, which jointly models textual instructions as semantic intent and sparse visual instructions as spatial guidance, aiming to achieve precise and intent-faithful image manipulation. To this end, we first construct a textual-visual instruction paired dataset with more than 23K samples derived from dynamic videos, enabling aligned supervision for cross-modal instruction. We then propose TV-Edit, a Textual-Visual instruction unified Editing framework to contextualize drag or point-based visual instructions with image-text semantics and lift them into semantic-aware control representations for pretrained editing backbones. By integrating semantic intent and spatial constraints, TV-Edit leads to more precise spatial control, less instruction ambiguity, and stronger structural consistency than text-only or drag-based alternatives. Finally, we establish TV-Edit-Bench, a deliberately designed benchmark to evaluate semantic faithfulness, spatial alignment, and visual consistency with ground-truth references and controlled textual-visual variations for reliable assessment. Our experiments across multiple editing backbones demonstrate that TV-Edit consistently yields more precise and intent-faithful edits, significantly outperforming state-of-the-art instruction-based and drag-based baselines.

DepthMaster: Unified Monocular Depth Estimation for Perspective and Panoramic Images

While monocular depth estimation has achieved significant progress, achieving generalized metric depth estimation for both narrow field-of-view (FoV) perspectives and $360^\circ$ panoramas remains an unsolved challenge. Existing methods are often tailored to specific camera types and struggle to produce accurate metric depth that generalizes across diverse settings. This limitation stems from two key challenges: the inherent geometric discrepancy between perspective and panoramic cameras, and the scarcity of panoramic training data with metric annotations. In this work, we introduce DepthMaster, a unified metric depth estimation framework. Rather than employing specialized networks to learn spherical distortions, we reformulate the problem by decomposing panoramic images into overlapping perspective patches. Crucially, distinct from prior projection-based methods that rely on ad-hoc architectural modifications to handle boundaries, we introduce a novel Correspondence Consistency Loss (CCL) and inject virtual projection cameras as geometric priors, allowing us to seamlessly stitch the patches while avoiding specialized operators and keeping the backbone largely compatible with standard Transformer designs. This strategy also resolves the geometric differences by unifying all inputs into a canonical perspective representation, and effectively circumvents data scarcity by directly unlocking powerful metric priors from vast perspective datasets. Trained on a mixed dataset that contains only one panorama dataset, DepthMaster achieves state-of-the-art zero-shot performance on 13 diverse datasets, outperforming not only universal methods but also leading specialist models in both perspective and panoramic domains.

FitVTON: Fit-aware Virtual Try-On via Body-Garment Size Control

While diffusion-based virtual try-on has achieved impressive visual realism, most methods treat the task as 2D inpainting, prioritizing texture preservation over physical plausibility. Consequently, they often produce plausible-looking images that fail to reflect authentic garment fit across diverse body shapes. We present FitVTON, a Fit-aware virtual try-on model on different bodies in the wild. FitVTON encodes garment-body size through structured text prompts, and learn from simulated try-on triplets from parameterized garment model. To improve the fitting effects over garment silhouettes, we introduce two auxiliary head to predict the masks for both the garment and the exposed body. We further introduce a texture rectification stage to improve realistic appearance from simulated data. To evaluate the fitting fidelity, we curate a real-world dataset, FittingEffect3K, combining VLM-based scoring protocol. Both subjective and quantitive experiments show that FitVTON demonstrate authentic fitting fidelity, with significant sizing accuracy and shape preservation over state-of-the-art methods while maintaining competitive image quality. Project Page: https://zenoning.github.io/FitVTON/.

Hierarchical Certified Semantic Commitment for Byzantine-Resilient LLM-Agent Collaboration

Byzantine collaboration among large-language-model agents requires a finality-control primitive: given delivered stochastic, structured natural-language proposals, the protocol must decide whether the round supports a commit, what kind of commit, or a typed safe abort. Naive aggregation hides this choice behind a single verdict; classical Byzantine fault tolerance hides it behind byte-identity that LLM proposals do not satisfy. We introduce Hierarchical Certified Semantic Commitment (H-CSC), a BFT-inspired protocol that converts embedding-derived finality signals over verdict-conditioned proposal groups into one of three typed outcomes: a semantic_commit (a 2f+1 within-verdict semantic core backs the verdict, emitting a parameter-bound digest over the quantised aggregate), a verdict_commit (strong verdict margin but dispersed semantic rationale, emitting a verdict-level certificate without claiming a semantic aggregate), or an explicit abort with a typed reason. The contribution is typed finality, not raw commit accuracy. On a controlled semantic-poisoning diagnostic (BCS_v1, 120 episodes), H-CSC commits with low angular deviation on BFT-feasible buckets (0.31 to 2.04 degrees) and aborts 100% of beyond-BFT rounds (n<3f+1) as intended. On a real LLM-agent claim-verification benchmark (MVR-50, 50 tasks) under paired static and rushing Byzantine attacks, H-CSC commits 0.90/0.92 with honest-reference-invalid rates of 0.02/0.00, statistically matching a strong certificate-emitting verdict-only baseline. Unlike that baseline, H-CSC also emits an embedding-backed semantic_commit digest on 74%/72% of rounds, supplying typed provenance. A strict-semantic ablation commits only 0.54/0.48, showing the verdict-level fallback is necessary for coverage (+0.36/+0.44) at the same <=0.04 safety floor; a 100-task cross-model check across four LLMs preserves invalid_hmaj within 0.00 to 0.03.

Masking Stale Observations Helps Search Agents -- Until It Doesn't: A Regime Map and Its Mechanism

Long-horizon search agents accumulate large amounts of retrieved content across many tool calls, making context-budget efficiency increasingly important. A minimal intervention is to mask stale observations from the context as the trajectory progresses, but it remains unclear when this form of context management helps and why. We study observation masking through a systematic sweep over various agent backbones (4B to 284B parameters) and three retrievers on offline and live-web agentic search benchmarks. We find that the accuracy gain from masking follows an asymmetric inverted-U shape when plotted against the model's accuracy without context management: a plateau under weak retrievers, a peak when a strong retriever meets a mid-capacity model, and a sharp collapse when the model is saturated. This pattern reflects the interaction between retriever recall and the model's implicit filtering capacity, rather than either factor in isolation. Mechanistically, masking implements a token-for-turn trade-off: it removes observations the model has largely stopped attending to and pages the agent rarely re-opens. The added turns help when they convert failures into successes, but they fail when masking removes evidence the model would otherwise have used. We therefore reframe context management as a regime-dependent intervention and provide a holistic perspective for analyzing context use in agentic deep search. We release our scaffold and trajectories here (https://github.com/i-DeepSearch/observation-masking) to support future research.