#01Jul 27, 2026
cs.CV
Infrared Imaging Empowered by Artificial Intelligence for Pediatric Skeletal Triage: A Narrative Review and Future Perspectives
Sajad Amiri, Pardis Afshar, Elham Anjomshoa
Background. Pediatric musculoskeletal trauma represents up to 18% of pediatric ED visits, yet diagnosis still depends on ionizing radiography. Cumulative low-dose radiation in early life raises lifetime leukemia and brain malignancy risk, motivating radiation-free triage alternatives. Objective. To synthesize evidence for a hybrid framework coupling broad-spectrum infrared (IR) imaging with deep-learning cross-modal translation to generate clinically interpretable synthetic-radiograph reconstructions from non-ionizing data. Approach. We review five IR spectral windows spanning 650 nm to 1 mm - NIR-I, NIR-II, SWIR, MIR/LWIR, and THz - and how dual-geometry (transmission/reflection) acquisition exploits wavelength-specific tissue depth and biochemical sensitivity. We summarize image-to-image translation networks (Pix2Pix, CycleGAN, Swin-Unet) and feature-matching algorithms (SuperPoint, SuperGlue, ALIKED, LightGlue) used to align and fuse IR data into radiograph-equivalent reconstructions. Implications. Pediatric anatomy - smaller cross-sections, thinner cortical bone - favors IR penetration, enabling compact, portable, non-ionizing triage hardware. Feasibility is grounded in fNIRS and transcranial photobiomodulation evidence: near-infrared light passes through skin, skull, and cortex with sufficient signal for hemodynamic monitoring - a longer, more attenuating path than through a pediatric forearm or distal leg. Key barriers: paired IR/X-ray dataset construction, AI-as-medical-device regulatory pathways, generalization across body habitus and skin pigmentation, and acquisition-protocol standardization. Conclusions. Integrated multi-spectral IR+AI imaging is a promising radiation-free complement to pediatric skeletal radiography. Progress requires multi-center paired datasets, externally validated models, and IR source safety qualification under IEC 60825-1.
#02Jul 27, 2026
cs.CV
Evidence Attribution in Visual Document Understanding without Coordinates or Region Labels
Zhuchenyang Liu, Yao Zhang, Yu Xiao
Reliable visual document understanding requires a model to attribute each answer to the evidence regions that support it. Recent benchmarks and systems express this step through a coordinate interface: the model outputs the coordinates of bounding boxes that mark the evidence regions in the document. Under this interface, vision-language models often fail to identify the right regions even when the answer is correct, a failure known as Attribution Hallucination. We present a study that investigates whether this failure is partially limited by what the model can express through coordinates. On a verified bilingual CiteVQA subset, we compare the coordinate interface with a language interface in which the model outputs only text, quoting its evidence verbatim, and a multimodal retriever returns the location of each quote as a page region proposed by a layout parser (tables and figures are quoted through their captions or notes); the comparison is repeated over six open vision-language models. Compared with the coordinate interface, evidence recall rises from at most 8 points to between 26 and 47 and the hallucination rate roughly halves, with little change in answer quality. Building on this comparison, we use the same quote-and-retrieve pipeline as a training scaffold: because region-level evidence labels are expensive to collect for long documents, we introduce a GRPO recipe whose reward is a judge's reading of the gold answer and crops of the retrieved regions, training the model to quote better evidence without any region labels and raising an 8B backbone's strict attributed accuracy from 22.4 to 33.8. These findings indicate a practical path to improve attribution"without a coordinate interface and without costly region-level supervision.
#03Jul 27, 2026
cs.CV
DecoupleMix: Decoupled Ratio Search and Convex Allocation for Scalable VLM Data Recipes
Jiahao Xie, Zhongbin Guo, Qianle Wang and 4 more
While data curation for Vision Language Models (VLMs) is increasingly active, public practice for constructing pretraining mixtures remains largely heuristic: practitioners stack datasets that pass quality filters, set cross-domain ratios by intuition, and lack a principled, attributable criterion for admitting new data, while frontier recipes remain undisclosed. We formulate data construction as a systematic mixture-optimization problem and turn it into a reproducible engineering discipline by decoupling the mixture into two orthogonal sub-problems: inter-class ratios across capabilities and intra-class ratios within a category. For inter-class allocation, we use a single-variable iterative search; for intra-class composition, we apply a multidimensional, dataset-level assessment scoring Quality and Difficulty, and formulate selection as a constrained convex optimization with a diversity objective. The DecoupleMix framework delivers two critical capabilities: guiding what data to collect next and rendering dataset validation a controlled, attributable experiment. Experiments show our approach consistently surpasses heuristic baselines. Moreover, optimal ratios discovered on small-scale proxies transfer seamlessly to larger scales without retuning. Using 80B additional multimodal continue-pretraining tokens, our VLM is competitive with strong open-source models trained with substantially larger multimodal budgets.
#04Jul 27, 2026
cs.CV
Rethinking Expert Training for Model Merging with Prompt Learning
Christos Georgakilas, Aniello Panariello, Samir El Karrat Moreno and 3 more
Model merging aims to combine multiple domain-specialized experts trained from a shared foundation model into a single multi-task model. Existing approaches largely focus on improving the merging procedure itself and typically assume experts obtained through full-parameter fine-tuning. In this work, we revisit expert training for model merging. We first show that prompt-based adaptation provides a strong baseline: independently learned prompts can be exploited across tasks while keeping the backbone fixed, avoiding the interference introduced by weight merging. Building on this observation, we introduce Dual-Tuned Experts (DTEs), a two-stage training strategy that first learns prompts and then fine-tunes the vision encoder. This reduces the magnitude of task-specific parameter updates and produces experts with higher merge compatibility. Experiments across multiple CLIP architectures, full fine-tuning, and LoRA experts show that DTEs consistently improve merged performance of standard merging approaches and remain effective even when combining heterogeneous sets of experts.
#05Jul 27, 2026
cs.CV
Rethinking Classifier-Free Guidance in On-Policy Diffusion Distillation
Bingnan Li, Haozhe Wang, Haozhong Xiong and 5 more
On-policy distillation (OPD) adapts diffusion models by querying a teacher along trajectories generated by the current student, but how it should behave under classifier-free guidance (CFG), a default component of modern diffusion systems, remains poorly understood. Existing OPD methods naturally extend velocity matching to the CFG-composed prediction, directly matching teacher and student guided velocities. We show that this objective is under-identified at the branch level: positive- and negative-branch errors can compensate in the guided prediction. Through two contrasting cases, we find that naive matching remains effective under shared negative conditioning, where both branch errors decrease jointly. When the model's native CFG schema retains privileged information in the teacher's negative branch that is unavailable to the student, however, this joint reduction breaks down and the composed objective induces antagonistic branch-error dynamics, reducing the positive-branch error while increasing the negative-branch error. We term this failure mode Negative Branch Asymmetry (NBA). To address NBA, we introduce Positive--Direction Matching (PDM), a branch-aware OPD objective that separately constrains the positive prediction and the CFG conditional direction. We apply PDM to dense-to-sparse video control, where naive guided matching is highly sensitive to inference guidance scales, while branch-aware supervision enables more robust and effective knowledge transfer.