Search arXiv⌕ Search

arXiv subjects

Bo Ma

Publications and source records attributed to Bo Ma.

At least 19 recordsLinked to original sources
arXiv2026

JDCNet Uses Confidence Gated Privileged Modality Distillation for Cost Preserving X Ray Inference

Training-only privileged-modality learning uses an expensive modality during training while keeping a single-modality deployment path. We study chest radiograph (X-ray) classification with same-patient computed tomography (CT) available only during training. JDCNet adds an auxiliary CT-teacher target, the teacher's argmax label or its softmax distribution through a Kullback--Leibler term (soft-KL), only on samples whose teacher confidence exceeds a threshold; the deployed model has the same ResNet-18 computation graph as the supervised baseline. On 510 same-patient pairs from the public BIMCV-COVID19 releases of the Valencian Region Medical ImageBank (113 COVID-19, 397 non-COVID), with patient-level five-fold cross-validation and three seeds, 15 of 16 adaptively explored JDCNet configurations improved mean balanced accuracy (BA) over the supervised baseline (average $+0.020$), but none remained significant after Holm adjustment. In a pre-registered experiment with 240 batch-matched supervised restarts on the same splits, batch size had no detectable effect on the baseline (BA $0.620$ versus $0.625$), and when both methods selected one of 16 runs per cell on the validation partition, JDCNet exceeded supervised training by $+0.037$ BA (95\% confidence interval $+0.016$ to $+0.058$; patient-level $+0.007$ to $+0.068$). Logit distillation and contrastive alignment did not help. Teacher diagnostics indicate that the gate acts mainly as CT-informed sample weighting. All models, including JDCNet, performed at chance on an external radiograph set. Confidence-gated CT supervision thus gives a small within-cohort advantage on one public cohort and a binary task; generalization requires independent paired replication.

cs.CV↗
arXiv2026

Auditing Targetless Camera-LiDAR Extrinsic Correction before Recursive Use

Camera-LiDAR fusion repeatedly uses an extrinsic transform in projection, association, and recursive state estimation. We study whether a targetless correction should be admitted to a downstream consumer only after separate audits of correction quality, confidence ranking, and recursive risk. Our log-disjoint protocol evaluates these stages under synthetic extrinsic drift on Argoverse 2 and treats cross-dataset and odometry experiments as bounded diagnostics. On 27 test-split logs, single-frame Canny alignment is harmful in the mild tier. Multi-frame alignment improves severe-tier rotation by 2.52 degrees but worsens translation by 0.177 m; its declared joint error-magnitude improvement is also negative (-0.158, 95% CI [-0.223, -0.098]), so the rotation gain is not a six-degree-of-freedom correction success. A zero-injection check moves away from the true extrinsic, and solver-disagreement AUROC on selected splits does not yield a validated nonzero-coverage risk-coverage operating point. In a 15-log final system role, the locked conservative policy admits no candidate and equals Raw, whereas Always-correct increases trajectory error. These results support an auditable separation of correction, confidence, and recursive contamination; they do not establish a deployable calibration gate, positive odometry or SLAM improvement, map consistency, or closed-loop safety.

cs.CV↗
arXiv2026

Resolving the Complex Line-Locking Structure in Quasar J004613+010425 with VLT/X-shooter Spectroscopy

Line-locking (LL) signatures in quasar absorption-line systems provide direct observational evidence for radiatively driven outflows. We present a detailed investigation of LL phenomena in the quasar SDSS J004613.54+010425.7 ($z_{\rm em} = 2.150$) using medium-resolution VLT/X-shooter spectra, as part of the TOLL project. Compared with previous studies based on Sloan Digital Sky Survey spectroscopy, the X-shooter data resolve the absorption complex into twelve narrow absorption line (NAL) components spanning $\sim$$2700$ km/s in the C IV, N V, Si IV, and Ly$α$ transitions, a factor of $\sim$2.4 more components than identified from lower resolution spectra. We detect fourteen distinct LL pairs across three ionic species, forming one of the largest line-locking webs known to date: three Si IV pairs connecting components (2, 10), (4, 11), and (5, 12); six N V pairs organized into four groups; and five C IV pairs organized into three groups. Monte Carlo simulations show that the probability of this twelve-component LL web arising by chance is $< 10^{-5}$. Photoionization modeling with Cloudy yields ionization parameters of $\log U \approx -2.4$ to $-0.6$ and total hydrogen column densities of $\log N({\rm H}) \approx 18.4$-$20.6$ for the twelve components. A comparison of the properties of shadowed and shadowing clouds shows that ion column densities and ionization parameters typically agree within a factor of $\sim$3, broadly supporting the standard radiative acceleration picture, although several outliers suggest that multi-ion locking and cloud-cloud interactions may also regulate LL stability. These results demonstrate that medium- and high-resolution spectroscopy is essential for resolving the full complexity of line-locked quasar outflows.

astro-ph.GA↗
arXiv2026

Mock Observations for the CSST Mission: CPI-C -- Targets for High Contrast Imaging

We introduce CPISM, a simulation program developed for the Cool Planet Imaging Coronagraph (CPI-C) on the China Space Station Telescope (CSST). CPISM supports high-contrast exoplanet imaging by simulating observational conditions and instrumental effects to optimize target selection and observation strategies. The modular design includes target modeling, imaging simulation, observational effects, detector response, and data product generation modules, enabling flexible and realistic synthetic observations. Validation through simulations of a bright star shows strong agreement with theoretical expectations, confirming the program's accuracy. CPISM's modular design allows flexibility, accommodating different stellar and planetary models, and can simulate instrumental noise, cosmic rays, and other observational effects. This tool aids in data processing, signal-to-noise ratio analysis, and high-contrast photometry, contributing to future exoplanet discovery and characterization efforts. The program's outputs will enhance observation planning and scientific return for the CPI-C mission, providing critical insights into exoplanetary systems.

astro-ph.IM↗
arXiv2026

ExoEcho: Exoplanet Ephemerides Change Observations. III. A JWST Survey for Transit Timing Variations of 37 Exoplanets

Transit timing variations (TTVs) provide a powerful probe of planetary dynamics and orbital evolution. We extend the ExoEcho project by incorporating James Webb Space Telescope (JWST) transit observations and analyze 84 light curves of 37 exoplanets, including 20 hot Jupiters, 11 Neptune-like planets, and 6 super-Earths. We derive updated transit ephemerides and conduct a systematic search for long-term and periodic TTV signals. We identify two strong TTV candidates, including HATS-6~b and HAT-P-18~b. The hot Jupiter HATS-6~b exhibits a long-term trend with $\dot{P} = 230.5 \pm 46.8\ \mathrm{ms\ yr^{-1}}$, potentially driven by tidal interactions with its host star. In addition, the Neptune-like planet HAT-P-18~b shows periodic TTV signal with amplitude of $92.9 \pm 13.7$~s. We also show that the previously claimed positive period derivative in WASP-121~b is inconsistent with our new transit timing data. Although continued monitoring is required to confirm these findings, our results demonstrate the potential of JWST for high-precision transit timing and extend TTV studies beyond the traditional hot-Jupiter population to lower-mass exoplanets.

astro-ph.EP↗
arXiv2026

Composition without a Host: Carrying the Spatiotemporal-Composability Guarantee of Plugin-Based Agents across Processes

Plugin-based agents assemble capabilities at runtime, and the spatiotemporal-composability calculus proves a reversibility guarantee for this assembly. However, the guarantee is carried by a single process, which confines all components, sessions, and recovery records to one failure domain, where a fault spreads past the plugin boundary, and process death interrupts every session the process hosts. Resting only on the hypotheses the calculus already states and the stateless interface of the model call, this paper relaxes the single-process restriction of the calculus to an arbitrary assignment of components and records to processes, gives four sufficient conditions, and proves with Theorem 1, derived from the four lemmas, that the reversibility guarantee holds across processes when these conditions are met. Based on Theorem 1, this paper constructs Logos, a cross-process plugin-based agent in the peer-process and name-routed form of ROS, where a plugin is a process, the router holds only a rebuildable routing table, and the session state needed for recovery lives in an append-only transcript owned by no process. Under one fault on two hundred benchmark tasks across three configurations, the single-process reference lost every session and scored 1.5 percent on the official validator, the MCP configuration kept its sessions while spending 1099 calls on a dead endpoint, and Logos kept every session alive, wasted zero calls, and succeeded on 120 tasks against 102 for both configurations combined. At the mechanism level, eighty sessions terminated at four points of the tool-call cycle all resumed with no repeated action, 3,500 concurrent calls paired with zero violations, and one bus hop cost 1 in 823 of the model's first token. The results show that the reversibility guarantee holds across processes and that assembly itself can leave the host process.

cs.AI↗
arXiv2026

PropGen: Automated Property Generation for Property-Based Testing of Mobile Apps

Mobile apps often suffer from functional bugs that do not cause crashes but instead manifest as incorrect behaviors under specific user interactions. Such bugs are difficult to detect by conventional automatic testing techniques because they often lack explicit \textit{test oracles}. Property-based testing can effectively expose them by specifying intended behavior as properties and checking them under diverse interactions. However, its practical use is limited by the reliance on manually written properties, which are difficult and expensive to construct. To address this limitation, this paper explores the use of large language models (LLMs) to automate property construction for property-based testing of mobile apps. This is challenging in two ways. \textit{First}, it is difficult to systematically uncover and execute diverse app functionalities. \textit{Second}, it is difficult to derive valid properties from functionality execution results. To address these challenges, we introduce PropGen, which infers candidate app functionalities as hypotheses from GUI states, validates each hypothesis by executing it to collect behavioral evidence, synthesizes properties from the collected evidence, and refines imprecise properties based on testing feedback. We implemented PropGen and evaluated it on 12 real-world Android apps. The results show that PropGen can effectively identify and execute app functionalities, generate valid properties, and refine most imprecise ones. Across all apps, PropGen inferred 1,210 valid functionalities and correctly executed 977 of them, compared with 491 and 187 for the baseline. It generated 985 properties, 912 of which were valid, and successfully refined 118 of 127 imprecise ones exposed during testing. Using the resulting properties, we found 25 previously unknown functional bugs, many of which were missed by existing testing techniques.

cs.SE↗
arXiv2026

Irradiated Atmospheres VII. Effect of Mixing Energy Flux on Atmosphere-Interior Coupling

Highly irradiated, underdense rocky planets such as puffy Venuses may host strong coupling between their atmospheres and magma-ocean interiors. Previous studies have mainly treated this coupling through radiative transfer and chemical equilibrium, leaving the role of vertical mixing unclear. Here we quantify how energy transport associated with vertical mixing modifies the thermal structure, volatile chemistry, and emission spectra of puffy Venuses. Atmospheric circulation and gravity-wave breaking are possible dynamical sources of such transport. We show that this energy transport provides an additional greenhouse-like heating pathway by warming the deep atmosphere, thereby increasing the pressure and temperature at the magma-ocean surface. The resulting thermal changes also reshape the atmospheric composition, especially the abundances of \ce{CO} and \ce{H2O}, and change the partitioning of C-, H-, and O-bearing volatiles between the atmosphere and the magma ocean. Moreover, these changes reshape the mid-infrared planet--star flux ratio, especially at low equilibrium temperatures and in higher-mass planets. Under high C/O, it slightly increases the flux near $3.5$--$4.0\,μ$m, but more broadly reduces atmospheric transparency and suppresses the flux. This may affect radius estimates for puffy Venuses over a limited temperature range.

astro-ph.EP↗
arXiv2026

Mass-Orbital Period Distribution of Massive White Dwarfs Formed Through Stable Mass Transfer

White dwarfs (WDs) in binaries can form through either the stable mass-transfer process or common envelope evolution (CEE). Compared to CEE, the stable mass-transfer process can lead to a distinct mass$-$orbital period ($M_{ \mathrm{WD}}$-$P_{ \mathrm{orb}}$) relation. Thus, this relation of WDs contains the information about the evolution channels. We can study the relation in WD binary systems to determine whether their progenitors undergo a CEE. We use the stellar evolution code MESA as our primary computational tool and adopt the quasi-adiabatic criterion to ensure that our models satisfy the conditions for stable mass transfer. Our study considers different mass-transfer schemes, varying metallicities, and the relation for both low-mass and intermediate-mass progenitors. Previous studies have focused on the relation for low-mass progenitors, which cannot explain some long-period, high-mass WD binaries. Our results show that the $M_{ \mathrm{WD}}$-$P_{ \mathrm{orb}}$ distribution for intermediate-mass progenitors whose cores remain nondegenerate prior to central helium burning can account for the formation channels of long-period and massive WD binaries.

astro-ph.SR↗
arXiv2026

BodhiPromptShield: Pre-Inference Prompt Mediation for Surface-Form Privacy Propagation in LLM Agent Pipelines

In LLM agent pipelines, prompt privacy risk propagates beyond a single model call: raw user content enters retrieval queries, memory writes, tool arguments, OCR-derived text, and logs, and every downstream copy inherits what the first write contained. Existing de-identification pipelines protect document boundaries but not this cross-stage surface. We present BodhiPromptShield, a policy-aware mediation layer that detects sensitive spans before they propagate, replaces each with a typed placeholder, a semantic abstraction, or a secure symbolic token under a configured policy, and defers restoration to authorized execution boundaries. We evaluate it under one protocol against Presidio, Casper-style sanitization, an LLM sanitizer, and transformer and learned detectors, on 300 AI4Privacy documents, 493 PrivacyLens trajectories, 200 PrivacyLens tasks scored by that benchmark's own judge, and AgentDojo tasks under injection. Three findings result. Identifier propagation is controllable: residual exposure falls to 7.4% on AI4Privacy and 1.8% on PrivacyLens, and exact identifiers in an agent's final action fall from 13.7% to 2.1-3.1%. Restoration timing governs what every stage upstream of the authorized boundary sees: deferring it leaves 1.6% of protected values readable in the released context against 51.0%, and 2.7% against 4.8% in what the agent emits, for 0.11 helpfulness points. Measuring factual disclosure is harder: a word-overlap metric and an LLM judge both report that mediation leaves facts intact, and both disagree with blind human annotation (kappa = 0.25 and 0.09). The human labels reverse that: inferability falls from 100% to 24-53% under mediation, so semantic-leakage measures need human validation before they are trusted. These are systems results on English text with open-weight models, not formal guarantees.

cs.CR↗
arXiv2026

Reading Decoder Trajectories: Training-Free Counterfactual Query-Trajectory Reliability for Small-Object Detection

Small-object detection remains challenging because limited pixels cause information loss and suppress the scale knowledge encoded in pretrained detectors. Existing approaches mainly improve representations through multiscale training, architecture redesign, or parameter adaptation, implicitly assuming that frozen models lack the required capability. We challenge this assumption and hypothesize that small-object knowledge already exists in frozen detectors but remains underactivated and unstable during query evolution. To test this hypothesis, we propose Counterfactual Query-Trajectory Reliability (CQTR), a training-free framework that elicits latent responses through counterfactual scale interventions and interprets candidate reliability from decoder-internal spatial convergence, semantic persistence, and cross-scale conflicts. A small unlabeled training subset selects the appropriate correction mechanism for each model-data stream, without parameter updates or target-domain annotations. Across 27 combinations of nine frozen detectors and three datasets, CQTR consistently improves average precision (AP) and average precision for small objects (APs). Closed-loop analyses further show that scale intervention activates latent responses, trajectory evidence predicts ground-truth support, and unlabeled routing selects the more effective branch. CQTR therefore reframes small-object detection from external scale augmentation to the activation and reliability assessment of latent scale knowledge.

cs.CV↗
arXiv2026

CODE: Cross-Modal Calibration and Dynamic Suppression for Open World Object Detection

Open World Object Detection (OWOD) built on multimodal foundation models often suffers from semantic ambiguity caused by unidirectional text-to-vision matching, while rigid outlier penalties may over-suppress unknown objects near known-class decision boundaries. We propose CODE (Cross-Modal Calibration and Dynamic Suppression), a unified inference-time framework with three complementary components. Cross-Modal Joint Confidence Calibration injects global visual prototypes to calibrate text-driven known-class predictions. Uncertainty-Guided Universal Objectness Enhancement measures classification hesitation from local visual responses to strengthen potential unknown objects. Dynamic Outlier Suppression via Confidence Margin replaces rigid suppression with a margin-aware adjustment that preserves ambiguous out-of-distribution instances. Experiments on the Real-World Detection benchmark demonstrate that, with the OWL-ViT L/14 backbone, CODE achieves 21.7 U-mAP and 40.8 K-mAP in Task 1, surpassing the previous state of the art by 2.6 and 2.3 points, respectively.

cs.CV↗
arXiv2026

Where Grounding Accuracy Lives on the IoU Curve: Label-Free Inference-Time Boundary Refinement

Vision--language models can identify the correct referent while returning an imprecise bounding box. We study whether a frozen direct-answer model can use its own prediction to allocate one additional localized observation without accessing target annotations at inference. Label-free precision refinement (LFPR) routes predicted-small regions to a higher-resolution pass, re-grounds the expression inside a context crop, admits a candidate only under fixed geometric guards, and returns a fixed coordinate-wise midpoint. We report results across three evidence tiers. On 31,921 retrospective Ref-L4 expressions, LFPR raises mAcc$_{0.5:0.95}$ from 72.947\% to 76.013\% (Acc@0.5 88.531\%$\to$89.725\%, Acc@0.9 55.788\%$\to$61.142\%). A frozen transfer to 30,969 RefCOCO/RefCOCO+/RefCOCOg expressions improves every dataset at Acc@0.5, mAcc, and mean IoU (pooled mAcc $+0.645$, Acc@0.5 $+0.817$), while Acc@0.9 is unchanged overall: routing alone gains $+1.162$ points there, but crop, guards, and fusion give back $-1.192$, offsetting rather than showing no strict-IoU effect. A prospective, image-disjoint Flickr30K Entities evaluation improves every endpoint (mAcc $+0.973$, Acc@0.9 $+1.022$), more strongly under a single-box variant (mAcc $+2.575$, Acc@0.9 $+3.689$). The same operator applied to two released grounding specialists improves every endpoint (Acc@0.9 $+1.569$/$+6.716$ for EGM-4B/8B) at roughly twice the latency, composing with specialist training rather than replacing it. A genuine unguarded control (guard removed from the same candidates) underperforms the incumbent on every metric, showing the guard is load-bearing. Together, these results show that referent selection and boundary precision are partially separable, with different components moving opposing regions of the IoU curve -- behavior a single threshold cannot reveal.

cs.CV↗
arXiv2026

A Systematic Gaia--ZTF Search for Short-Period Blue Compact-Binary Candidates

We present a catalog of 147 short-period (10.34--106.46~min) blue compact-binary candidates, identified by combining Gaia DR3 astrometry and photometry with ZTF DR23 light curves via a Gaia selection, period searches, and machine-learning morphology ranking. Of these, 111 lack prior compact-binary classifications. Multiwavelength data (DESI DR1, GALEX, AllWISE) reveal a heterogeneous sample: on the Gaia colour--magnitude diagram, 52 sources lie on the white-dwarf locus, 69 in the hot-subdwarf region, and 26 are intermediate. Among 26 sources with DESI spectra, only about one third follow the white-dwarf cooling sequence; the rest are more luminous blue stars with white-dwarf-like low-resolution spectra. We highlight a prioritized subset of new white-dwarf-locus candidates for follow-up, including ten with periods below 40~min and none with existing radial-velocity data. Under fiducial binary assumptions, 17 of these newly identified white-dwarf-locus candidates would exceed the adopted LISA signal-to-noise threshold (led by a 37~pc white dwarf), with the count depending on chirp mass (9 for $0.15\,M_\odot$, 17 for $0.3\,M_\odot$, 21 for $0.6\,M_\odot$), assuming orbital modulation. However, for most of the white-dwarf-locus sample, observed modulation amplitudes exceed any plausible ellipsoidal signal by three to five orders of magnitude, implying that rotating magnetic or chemically inhomogeneous single white dwarfs offer a viable alternative that ZTF photometry alone cannot rule out---the catalog includes at least one confirmed case. We release the full 147-source catalog, including periods, Gaia/spectroscopic classifications, harmonic/ellipsoidal diagnostics, and supplementary tables of fiducial GW estimates and UV--IR photometry.

astro-ph.SR↗
arXiv2026

Binary Constraints on the Origin of Nitrogen-rich Field Stars

Recent JWST observations have revealed galaxies with unusually high N/O ratios, suggesting that nitrogen enrichment may be common in intense star-forming environments in the early Universe. In the Milky Way, nitrogen-rich(N-rich) stars in the Galactic field have long served as probes of early Galaxy formation and globular cluster enrichment. However, the identification of binaries among these stars raises the possibility that binary mass transfer could contribute to their origin. In this work, we utilize multi-epoch radial velocities and element abundances from APOGEE DR17 to constrain their formation sites. Among 266 N-rich field stars, 33 exhibit radial velocity variations of $Δ{\rm RV} > 1\,{\rm km/s}$, including 10 robust spectroscopic binaries identified using the $F_2$ statistic within a well-sampled subset of 46 stars. The resulting close-binary fraction ($21.7\pm6.1\%$) is statistically indistinguishable from that of chemically normal field stars ($18.1\pm0.6\%$), showing no evidence of the excess expected from AGB binary pollution. This is further supported by the absence of correlation between [N/Fe] and [Ce/Fe] and the lack of [C/Fe] enhancement. Crucially, we detect an anti-correlation between binary fraction and [Al/Fe], with strongly Al-enhanced stars ($[\mathrm{Al/Fe}] \gtrsim 0.5$) exhibiting a reduced binary fraction ($< 10\%$). This trend serves as a dynamical fingerprint of high-density environments, consistent with the efficient disruption of binaries via three-body interactions in GC cores. Our results do not support binary mass transfer as the dominant formation channel for N-rich field stars; they are predominantly GC escapees that retain the dynamical memory of their dense birth sites.

astro-ph.GA↗
arXiv2026

Artificial Intelligence for Spatially Reconfigurable Antennas: Movable, Fluid, and Pinching Antenna Systems

Recently, sixth-generation (6G) wireless networks have moved beyond fixed-array designs toward antenna architectures that can adapt their spatial configuration to specific environmental conditions. Movable antenna, fluid antenna, and pinching antenna systems represent this principle in different ways, but they share a common vision: exploiting spatial flexibility as an additional degree of freedom (DoF) to improve communication, sensing, security, and resource efficiency. These new techniques, however, also bring challenging problems, as antenna configuration must be jointly considered with channel acquisition, beamforming, mobility, and network resource management. Therefore, artificial intelligence (AI) has become an important tool for learning fast and adaptive control policies for these highly coupled systems. In this survey, we provide a unified review of AI for spatially reconfigurable antenna systems. We first introduce the basic principles of movable, fluid, and pinching antennas, which is followed by a summary of the latest AI-enabled designs according to their primary optimization objectives. Furthermore, we compare the roles of deep learning (DL), deep reinforcement learning (DRL), multi-agent reinforcement learning (MARL), graph learning, Transformers, large language models (LLMs), and structure-guided learning across different antenna architectures. Finally, we discuss open challenges and future directions toward scalable, robust, and hardware-aware intelligent reconfigurable antenna networks.

eess.SP↗
arXiv2026

Clean-Reference Streaming Detection of Lens Occlusion and Photometric Transitions for Camera Tamper Monitoring

A surveillance camera is an image sensor whose silent physical degradation invalidates every downstream consumer of its data. In-situ integrity alarms for such vision sensors require low false-alarm rates, bounded computation, and diagnosable behavior under nuisance illumination changes. This paper studies a deliberately narrow streaming integrity monitor for two low-cost sensor-fault signatures: texture-collapsing lens occlusion and abrupt photometric scene transition. The detector compares sampled luminance and local-gradient statistics with a clean-only sliding reference, applies coarse-grid structured-light rejection and mode/rapid-brightness suppression, and emits at most one notification per tamper episode. We formalize the decision predicates and derive a consistency rule for when rapid-brightness suppression makes the scene-transition path unreachable. On 320 in-scope controlled sequences, the default state machine attains 0.800 F1 and 0.822 balanced accuracy (significantly better paired correctness than the strongest baseline, though the F1 margin is not statistically resolved); on a magnitude-swept public audit it attains the highest partial AUC under a 5\% false-alarm budget, and a separate extended-stress FPR-constrained sweep reaches 0.925 recall at 0.025 false-positive rate. Public Xiph, Bremen IoT, and UHCTD diagnostics show the fixed predicates preserve low false alarms while recall concentrates inside the declared envelope (UHCTD in-scope covered recall 0.667 versus 0.016 out of scope), and a 9.09-camera-hour verified-negative public audit records zero false alarms. The method is best interpreted as an auditable sensor-health subsystem rather than a universal camera-tamper classifier.

cs.CV↗
arXiv2026

HKVLM: Faithful Query--Region Binding for Frozen-Detector Visual Grounding

Visual grounding often fails even when the target object is present in the proposal pool, because the language-side referent is bound to the wrong region. We study this binding failure under frozen perception and ask whether an explicit query--region alignment hook, together with a perception-grounded abstention mechanism, can improve faithful grounding without retraining the detector or the vision-language backbone. HKVLM freezes a language-aligned open-vocabulary detector for localization and learns a lightweight hook that maps referential query embeddings to detector proposals in a shared space; a verifier abstains when no region sufficiently supports the query. We prove an exact proposal-level diagnostic decomposition, $(1-\mathrm{SeeErr})(1-\mathrm{SayErr})$, separating proposal-coverage failures from conditional binding failures, and a monotonicity result that characterizes the faithfulness--recall trade-off induced by abstention. Across RefCOCO, RefCOCO+, RefCOCOg, and POPE, HKVLM improves over untrained and trained matched-perception binding controls and substantially reduces hallucination through abstention. Strong coordinate-decoding and end-to-end fine-tuned baselines remain much higher in raw grounding accuracy, and a reasoning-stress set exposes binding as the main current bottleneck. We therefore present HKVLM as a diagnostic and mechanism-level study of query--region binding under frozen perception, not as an absolute localization leader.

cs.CV↗