#01Aug 18, 2026
cs.CV
PerFact: Perception-Derived Fact Prompting for 3D Brain MRI Report Generation
Jianyu Sun, Zhenxuan Zhang, Guang Yang and 1 more
Radiology report generation has matured almost entirely on 2D chest radiographs, where the default route to better reports is a larger backbone or a pre-training one on medical data. We revisit that assumption on 3D multi-sequence brain MRI, a volumetric multi-disease regime, and find that the model is not the lever. Zero-shot medical and radiology vision-language models transfer poorly to brain MRI, with chest radiograph specialists failing most conspicuously, and five backbones fine-tuned identically across three model families and an order of magnitude in scale differ only marginally. What determines the quality of the report is the information injected into the prompt. We delegate perception to upstream 3D segmentation and classification, serialize their outputs into a structured fact sentence, and prompt a LoRA-adapted vision-language model with it; we call this \textbf{PerFact}. In a controlled study that fixes the backbone, data split, target reports, and adaptation while varying only the injected grounding, perception-derived facts outperform retrieved prior reports, retrieval becomes redundant once facts are present, and end-to-end predicted facts remain effective without any ground-truth annotation at inference. The residual gap between predicted and oracle facts is explained by the granularity of the facts rather than by the generator. Closed-ended visual question answering comes at no measurable cost to report quality, though the grounding source has little effect on it. On 3D brain MRI, grounding information, not model choice, is the dominant controllable factor in report quality.
#02Aug 18, 2026
cs.CV
GenRec: Knowing Where to Reconstruct and Where to Generate
Ata Çelen, Jaewoo Jung, Federico Tombari and 4 more
Generative novel view synthesis from sparse input images is rarely all reconstruction or all generation: pixels visible in some source view have a unique correct value modulated only by view-dependent shading, while pixels in disocclusions or beyond the captured volume admit a distribution of plausible completions. Existing generative novel-view-synthesis methods conflate these regimes under a single uniform loss, blurring the line between geometric fidelity and creative hallucinations even when scene geometry is injected through warped point clouds or projected depth. We introduce GenRec, a multi-view flow matching model that builds the reconstruction--generation split directly into its architecture, supervision, and gradient flow. Guided by an observation mask derived from the source cameras and a monocular depth estimator, a flow matching backbone jointly denoises RGB and scene-coordinate maps across all target views, while a pixel-space refinement stage restores high-frequency detail on observed pixels; the same mask gates supervision so regression signals do not contaminate the generative prior. Across RealEstate10K, DL3DV-10K, and Mip-NeRF~360, in both single-view extrapolation and two-view interpolation, GenRec attains the best reconstruction fidelity in observed regions while also surpassing purely generative baselines on perceptual quality in unobserved ones, showing the effectiveness of our approach.
#03Aug 18, 2026
cs.CV
GS-Voxel: Fitting-Free Structured Latents for Large-Scale 3DGS Generation
Ming Qian, Zijian Wang, Minchao Sun and 5 more
Many scalable latent 3D generators operate on structured tensors, whereas pre-optimized 3D Gaussian Splatting (3DGS) reconstructions are unordered, spatially irregular, and vary widely in primitive count. We present GS-Voxel, a fitting-free structured latent framework, and evaluate it for large-scale aerial 3D Gaussian scene generation. GS-Voxel deterministically converts a compatible pre-optimized 3DGS reconstruction into sparse active voxels without additional per-scene optimization, retaining the sub-voxel positions and rendering attributes of the selected primitives. A GS-specific factorized VAE then separately encodes voxel geometry and local Gaussian attributes into sparse 3D latents whose size grows with the number of occupied voxels rather than being limited by a fixed scene-wide primitive count. We train image-conditioned flow models in the GS-Voxel latent space to generate aerial 3DGS scenes. A key application enabled by GS-Voxel is large-area scene generation: overlap-aware tiled inference extends synthesis beyond a single training crop conditioned on satellite-view images. Our results show that GS-Voxel provides structured latents for pre-optimized aerial 3DGS reconstructions, with latent capacity that grows with the number of occupied voxels.
#04Aug 18, 2026
cs.CV
Automated ACL Footprint Identification Using 3D Deep Learning
Ruida Cheng, Ali Uneri, Gabriel Gibson and 2 more
One of the most common reasons for anterior cruciate ligament (ACL) reconstruction failure is femoral tunnel malpositioning (ACL footprint center and tunnel orientation). Such failures may lead to the development of meniscal pathology and osteoarthritis. Accurate ACL femoral footprint identification is therefore essential for precise tunnel placement, restoration of the native knee joint mechanics, post-surgical knee joint health and prevention of graft failure. Recent advances in artificial intelligence (AI) bring new opportunities to improve image-guided orthopedic surgery. However, at present, existing AI research focuses primarily on ACL segmentation and rupture classification based on pre- and post-operative magnetic resonance (MR) images. Identification of the ACL footprint center using deep learning methods has not been thoroughly researched. Thus, the purpose of this study is to explore 3D deep learning models for ACL femoral footprint identification directly from 3D MR images. Two comprehensive 3D deep learning architectures were developed: a 3D graph convolutional neural network-based geometric model applied to 3D femoral meshes; and a 3D landmark-enhanced identification model based on 3D MR images. A total of 4883 right and 3087 left knee image sets were used from a publicly available database. Eighty percent (80%) were applied to model generation, and twenty percent (20%) were preserved for model testing. Both models achieved excellent performance; however, the image-based method outperformed the model-based method (average error of 2.1mm vs 2.8 mm). Thus, 3D deep learning provides a feasible clinical approach for ACL footprint localization and has the potential to improve ACL reconstruction footprint accuracy.
#05Aug 18, 2026
cs.CV
Scale Matters: Adaptive Granularity Selection for Cross-Species 3D Plant Organ Segmentation
Carla Salazar, Lazaros Nalpantidis
Recent 3D foundation models provide powerful feature representations for point cloud learning by controlling spatial granularity. However, relying on a fixed spatial granularity severely limits generalization in applications like plant phenotyping, where organ morphology and size vary substantially across species and growth stages. To address this, we propose AGS-PlantSeg, a few-shot 3D plant organ segmentation method that leverages the frozen Utonia (arXiv:2603.03283) foundation model combined with Adaptive Granularity Selection. By dynamically selecting the best granularity levels for each specific plant model, our method extracts optimized geometric features for a lightweight MLP segmentation head. Extensive experiments across PLANesT-3D (arXiv:2407.21150), Pheno4D , and Crops3D demonstrate that AGS-PlantSeg significantly improves cross-species generalization, achieving 88.9% average mIoU performance and outperforming fixed-granularity baselines by 2.5 mIoU points. Despite requiring minimal annotated data, our approach is highly competitive with fully supervised, plant-specific architectures.