#01Nov 2, 2026
cs.CV
SurgPace: Complexity-Aware Progressive Distillation for Surgical Multimodal Large Language Models
Zhicheng He, Xiaohan Xing, Hongqiu Wang and 7 more
Multimodal large language models (MLLMs) have shown strong potential in robotic surgery for question answering, scene understanding, and grounded assistance, but their high computational cost limits real clinical deployment. Knowledge distillation provides a practical way to build compact surgical MLLMs, yet direct distillation is not well suited to surgical data because surgical samples vary greatly in difficulty, visual ambiguity, and required clinical context, making uniform teacher-student transfer unstable. In addition, the role of supervision changes during training: early stages require stronger teacher guidance for multimodal alignment, while later stages rely more on task supervision for effective optimization. To address these challenges, we propose SurgPace, a complexity-aware continuous curriculum framework for progressive distillation of surgical vision-language models. SurgPace estimates sample complexity using predictive confidence and Lorentz geometric consistency, and uses the fused score to guide a continuous easy-to-hard training process. We further introduce hierarchical representation alignment, which aligns the student with the teacher at the feature, logit, and probability levels, while gradually shifting the training focus from teacher imitation to task supervision. This design separates sample-level curriculum from objective-level scheduling, leading to more stable and effective distillation. We evaluate SurgPace on Surg-396K, a benchmark built from four public surgical datasets and covering five dialogue paradigms. SurgPace consistently outperforms strong MLLM distillation baselines, demonstrating its effectiveness for building compact, memory-efficient surgical MLLMs with faster decoding under controlled benchmarks. Code is available at https://github.com/hezhicheng2002/SurgPace.
#02Nov 2, 2026
cs.CV
GPU-Native Real-Time Textured Mesh Reconstruction with Seam-Aware Multi-View Radiometric Correction
Hiroyuki Sato
We present a GPU-native volumetric-video system for real-time reconstruction of graphics-compatible textured triangle meshes from multiple RGB-D sensors. Existing TSDF systems accelerate fusion and mesh extraction, but extending them to per-frame textured meshing adds CPU mesh processing, CPU-side texturing work, or extra CPU-GPU transfers between stages. After input staging, our method keeps the reconstruction and texturing stages (TSDF fusion, Marching Cubes, connected-component noise removal, view-based texture mapping, inpainting, and radiometric seam correction) resident on the GPU; no intermediate mesh or image data returns to the host between these stages. A fixed-size-array representation of mesh topology gives constant-time neighbor access on the GPU and makes the downstream mesh-level operations data-parallel. Because we texture each face from a single camera, exposure seams that vary smoothly across each image remain; we estimate a per-camera low-frequency gain field and bake it into the atlas on the GPU. On the CMU Panoptic dataset, our method is faster than the baselines at every tested voxel resolution and number of sensors, achieves peak speedups of more than an order of magnitude over CUDA-accelerated nvblox and CPU-based Open3D, and approaches an offline textured-meshing baseline in perceptual re-rendering quality.
#03Nov 2, 2026
cs.CV
MuCHMDPoser: Multiview RGB-Camera Full-Body Pose Tracking for Virtual Reality
Antoine Maiorca, George Fletcher, Thierry Ravet and 1 more
Accurate full-body tracking in Virtual Reality (VR) is heavily underconstrained when relying solely on consumer-grade sparse sensors, such as Head-Mounted Displays (HMDs) and handheld controllers. Indeed, the self-avatar full-body pose must be inferred based on the 3 available upper-body joints tracked by these devices. This leads to significant tracking error, especially in lower-body pose. To overcome this issue, we propose to extend the tracking setup with external multiview RGB cameras. We show that integrating RGB-based pose estimation this framework is able to significantly decrease lower-body tracking errors, even under common artifacts in RGB-based estimates. Additionally, we evaluate a only-once HMD-camera calibration procedure that does not require any physical markers. We finally propose an implementation of the overall framework, achieving an end-to-end animation at 60 FPS on a Nvidia RTX3080 GPU. Codes are available at https://github.com/antmaio/MuCHMDPoser.
#04Nov 2, 2026
cs.CV
Balancing Representations by Identifying and Generating Under-represented Data
Filipe Laitenberger, Hirokatsu Kataoka, Yuki Asano and 2 more
Self-supervised learning inherits the coverage of its source data. When an unlabeled corpus is long-tailed, the representation becomes dense around frequent modes and thin where coverage is weak. Objectives for long-tailed SSL reweight the images already in the corpus and add no new ones. We introduce BRIDGE (Balancing Representations by Identifying and Generating Under-represented Data), a label-free feedback loop around the source set. The current encoder scores every source image by its mean kNN distance, and a budgeted acquisition rule selects images from a band of that score distribution where support is thin and the generator still produces faithful variants. A frozen image-to-image diffusion model then adds local variants of the selected anchors, and pretraining continues on the enlarged source set. We formulate acquisition as recoverability-weighted allocation of local support and derive when a finite budget should favor a sparse band that excludes the extreme tail and when reallocation after each repair is optimal under a fixed objective. Across five long-tailed, web-scraped and synthetic source regimes, BRIDGE has the highest seven-task linear-probe and kNN average of the compared methods, and paired experiments extend the gain to ViT-S under contrastive and self-distillation objectives, to low-label fine-tuning and to semantic segmentation. The linear-probe gain is smaller under masked reconstruction. Generated repair matches an oracle that restores the withheld real images, and under a frozen encoder the inserted images shrink the neighborhoods of the selected anchors about twice as much as those of matched unselected images. BRIDGE needs no labels and leaves the pretraining objective unchanged, so it can be added to pretraining on uncurated web data, where the encoder decides which images to generate.
#05Nov 2, 2026
cs.CV
RouteREPA: Routed Representation Alignment for Just Image Transformers
Yi Tan, Beier Zhu, Sirui Chen and 6 more
Representation alignment (REPA) is widely used to accelerate the training of latent-space diffusion transformers (DiTs). However, applying it to pixel-space models such as Just Image Transformers (JiT) is challenging. Our analysis reveals a pronounced representation conflict between discriminative alignment and pixel-space generation. Compared with its latent counterpart, JiT learns representations that are substantially less aligned with DINOv2. Applying REPA to JiT consistently reduces the reconstructability of its intermediate representations, whereas this effect is not observed in Latent JiT. We attribute this conflict to JiT directly reconstructing pixels, which requires its representations to hierarchically encode discriminative semantics and reconstruction-relevant visual information across network depths. To address these challenges, we propose Routed Representation Alignment (RouteREPA), which uses lightweight, representation-dependent routers to adaptively route task-oriented representations for alignment and generation. We further apply routing across multiple network depths to accommodate their depth-dependent representation requirements. Extensive experiments on ImageNet-1K at 256x256 resolution show that RouteREPA consistently improves JiT across model scales and training budgets, outperforming standard REPA. On JiT-H/16, RouteREPA further surpasses PixelREPA using only one-third of its training epochs without classifier-free guidance.