arXiv subjects

Shaotian Li

Publications and source records attributed to Shaotian Li.

3 recordsLinked to original sources

arXiv2026

CS-LDM: Clinically Guided Structure-Aware Latent Diffusion for Knee Radiographic Forecasting

Knee osteoarthritis (KOA) is a leading cause of physical disability and currently has no curative clinical treatment. Early diagnosis and accurate prognosis are therefore vital for managing its progression. Existing methods mainly predict future severity grades and provide limited visual explanation of structural change. Recent generation-guided approaches synthesize follow-up radiographs for progression assessment, but generated images are often judged by predictive accuracy or image realism, so they may satisfy a grading model while remaining morphologically unreliable. We therefore propose CS-LDM, a clinically guided, structure-aware latent diffusion framework that forecasts patient-specific future knee radiographs from the current X-ray and clinically available current KL grade, using LoRA adaptation, current-grade conditioning, and frozen-teacher diagnostic distillation. Experiments on the public Osteoarthritis Initiative (OAI) dataset show that CS-LDM improves future severity-grade consistency while maintaining competitive image fidelity. Further ordinal, OARSI, ROI-level, and stable/progressed analyses show that radiographic forecasting should be evaluated by balancing diagnostic consistency, anatomical plausibility, and structural preservation.

eess.IV↗

arXiv2026

FashionPose: Unified Text-Driven Fashion Synthesis with Joint Geometric and Photometric Control

Realistic and controllable garment synthesis is essential for fashion e-commerce, yet it demands precise coordination between human pose geometry and environmental photometry. Conventional pose-guided frameworks suffer from two fundamental limitations: they rely heavily on predefined skeletons from off-the-shelf estimators, restricting semantic flexibility; and they predominantly focus on studio-like generation under neutral lighting, failing to reconcile geometric configurations with complex, scene-specific illumination described in natural language. To bridge this gap, we propose FashionPose, a cascaded architecture that reconciles geometric and photometric control within a unified language-driven interface. Unlike conventional frameworks, our framework employs a decoupled yet synergistic strategy: (1) a bidirectional contrastive alignment mechanism that grounds textual semantics into an explicit geometric manifold, enabling template-free pose generation; (2) an identity-anchored synthesis module that translates these geometric priors into high-fidelity imagery while preserving fine-grained appearance; and (3) a prompt-conditioned relighting module that leverages the generated pose as a spatial anchor to achieve environment-aware shading. This hierarchical design effectively transforms high-level instructions into consistent visual representations, ensuring both structural precision and atmospheric harmony. To facilitate this paradigm, we construct PoseCap, a dataset with over 40,000 caption-keypoint pairs. Extensive experiments demonstrate that FashionPose outperforms existing benchmarks in pose accuracy and physical realism, providing a robust solution for personalized, scene-aware virtual fashion displays.

cs.CV↗

arXiv2026

Latent Anomaly Knowledge Excavation: Unveiling Sparse Sensitive Neurons in Vision-Language Models

Large-scale vision-language models (VLMs) exhibit remarkable zero-shot capabilities, yet the internal mechanisms driving their anomaly detection (AD) performance remain poorly understood. Current methods predominantly treat VLMs as black-box feature extractors, assuming that anomaly-specific knowledge must be acquired through external adapters or memory banks. In this paper, we challenge this assumption by arguing that anomaly knowledge is intrinsically embedded within pre-trained models but remains latent and under-activated. We hypothesize that this knowledge is concentrated within a sparse subset of anomaly-sensitive neurons. To validate this, we propose latent anomaly knowledge excavation (LAKE), a training-free framework that identifies and elicits these critical neuronal signals using only a minimal set of normal samples. By isolating these sensitive neurons, LAKE constructs a highly compact normality representation that integrates visual structural deviations with cross-modal semantic activations. Extensive experiments on industrial AD benchmarks demonstrate that LAKE achieves state-of-the-art performance while providing intrinsic, neuron-level interpretability. Ultimately, our work advocates for a paradigm shift: redefining anomaly detection as the targeted activation of latent pre-trained knowledge rather than the acquisition of a downstream task.

cs.CV↗