Zanxi Ruan

Qiuyu Kong, Shakiba Sharifi, Zanxi Ruan et al.

Is Segment Anything Model 3 (SAM3) capable in segmenting Any Pathology Images? Digital pathology segmentation spans tissue-level and nuclei-level scales, where traditional methods often suffer from high annotation costs and poor generalization. SAM3 introduces Promptable Concept Segmentation, offering a potential automated interface via text prompts. With this work, we propose a systematic evaluation protocol to explore the capability space of SAM3 in a structured manner. Specifically, we evaluate SAM3 under different supervision settings including zero-shot, few-shot, and supervised with varying prompting strategies. Our extensive evaluation on pathological datasets including NuInsSeg, PanNuke and GlaS, reveals that: 1.text-only prompts poorly activate nuclear concepts. 2.performance is highly sensitive to visual prompt types and budgets. 3.few-shot learning offers gains, but SAM3 lacks robustness against visual prompt noise. and 4.a significant gap persists between prompt-based usage and task-trained adapter-based reference. Our study delineates SAM3's boundaries in pathology image segmentation and provides practical guidance on the necessity of pathology domain adaptation.

Zanxi Ruan, Qiuyu Kong, Songqun Gao et al.

Edge-based representations are fundamental cues for visual understanding, a principle rooted in early vision research and still central today. We extend this principle to vision-language alignment, showing that isolating and aligning structural cues across modalities can greatly benefit fine-tuning on long, detail-rich captions, with a specific focus on improving cross-modal retrieval. We introduce StructXLIP, a fine-tuning alignment paradigm that extracts edge maps (e.g., Canny), treating them as proxies for the visual structure of an image, and filters the corresponding captions to emphasize structural cues, making them "structure-centric". Fine-tuning augments the standard alignment loss with three structure-centric losses: (i) aligning edge maps with structural text, (ii) matching local edge regions to textual chunks, and (iii) connecting edge maps to color images to prevent representation drift. From a theoretical standpoint, while standard CLIP maximizes the mutual information between visual and textual embeddings, StructXLIP additionally maximizes the mutual information between multimodal structural representations. This auxiliary optimization is intrinsically harder, guiding the model toward more robust and semantically stable minima, enhancing vision-language alignment. Beyond outperforming current competitors on cross-modal retrieval in both general and specialized domains, our method serves as a general boosting recipe that can be integrated into future approaches in a plug-and-play manner. Code and pretrained models are publicly available at: https://github.com/intelligolabs/StructXLIP.

Wanyin Cheng, Zanxi Ruan

Visual Question Answering (VQA) holds great potential for assisting Blind and Low Vision (BLV) users, yet real-world usage remains challenging. Due to visual impairments, BLV users often take blurry or poorly framed photos and face difficulty in articulating specific questions about what they cannot fully see. As a result, their visual questions are frequently ambiguous, and different users may interpret them in diverse ways. This leads to multiple valid answers, each grounded in different image regions-posing a mismatch with conventional VQA systems that assume a single answer and region. To bridge this gap, we present BLaVe-CoT, a VQA framework designed to reason about answer consistency in the face of ambiguity. Our method proposes diverse candidate answers using a LoRA-tuned BLIP-2 model, then grounds each answer spatially using PolyFormer, and finally applies a chain-of-thought reasoning module to assess whether the answers refer to the same or different regions. Evaluated on the VQA-AnswerTherapy benchmark, BLaVe-CoT outperforms previous methods and proves more robust to the ambiguity and visual noise common in assistive settings. This work highlights the need for VQA systems that can adapt to real human uncertainty and provide inclusive support for BLV users. To foster further research and accessibility applications, we have made the code publicly available at https://github.com/Accecwan/BLaVe-CoT.

Ziyue Liu, Davide Talon, Federico Girella et al.

Sketches offer designers a concise yet expressive medium for early-stage fashion ideation by specifying structure, silhouette, and spatial relationships, while textual descriptions complement sketches to convey material, color, and stylistic details. Effectively combining textual and visual modalities requires adherence to the sketch visual structure when leveraging the guidance of localized attributes from text. We present LOcalized Text and Sketch with multi-level guidance (LOTS), a framework that enhances fashion image generation by combining global sketch guidance with multiple localized sketch-text pairs. LOTS employs a Multi-level Conditioning Stage to independently encode local features within a shared latent space while maintaining global structural coordination. Then, the Diffusion Pair Guidance stage integrates both local and global conditioning via attention-based guidance within the diffusion model's multi-step denoising process. To validate our method, we develop Sketchy, the first fashion dataset where multiple text-sketch pairs are provided per image. Sketchy provides high-quality, clean sketches with a professional look and consistent structure. To assess robustness beyond this setting, we also include an "in the wild" split with non-expert sketches, featuring higher variability and imperfections. Experiments demonstrate that our method strengthens global structural adherence while leveraging richer localized semantic guidance, achieving improvement over state-of-the-art. The dataset, platform, and code are publicly available.

Muhammad Aqeel, Maham Nazir, Zanxi Ruan et al.

Medical image segmentation suffers from data scarcity, particularly in polyp detection where annotation requires specialized expertise. We present SynDiff, a framework combining text-guided synthetic data generation with efficient diffusion-based segmentation. Our approach employs latent diffusion models to generate clinically realistic synthetic polyps through text-conditioned inpainting, augmenting limited training data with semantically diverse samples. Unlike traditional diffusion methods requiring iterative denoising, we introduce direct latent estimation enabling single-step inference with T x computational speedup. On CVC-ClinicDB, SynDiff achieves 96.0% Dice and 92.9% IoU while maintaining real-time capability suitable for clinical deployment. The framework demonstrates that controlled synthetic augmentation improves segmentation robustness without distribution shift. SynDiff bridges the gap between data-hungry deep learning models and clinical constraints, offering an efficient solution for deployment in resourcelimited medical settings.