Tootfinder

CoMemo: LVLMs Need Image Context with Image Memory
Shi Liu, Weijie Su, Xizhou Zhu, Wenhai Wang, Jifeng Dai
https://arxiv.org/abs/2506.06279 https://…

CoMemo: LVLMs Need Image Context with Image Memory
Recent advancements in Large Vision-Language Models built upon Large Language Models have established aligning visual features with LLM representations as the dominant paradigm. However, inherited LLM architectural designs introduce suboptimal characteristics for multimodal processing. First, LVLMs exhibit a bimodal distribution in attention allocation, leading to the progressive neglect of middle visual content as context expands. Second, conventional positional encoding schemes fail to preser…

Anomalous Ionic Conductivity along the Coherent $\Sigma$3 Grain Boundary in ThO2
Miaomiao Jin, Jilang Miao, Marat Khafizov, Beihan Chen, Yongfeng Zhang, David H. Hurley
https://arxiv.org/abs/2507.04652

Anomalous Ionic Conductivity along the Coherent $Σ$3 Grain Boundary in ThO2
Understanding oxygen diffusion along grain boundaries (GBs) is critical for controlling ionic conductivity in oxide ceramics. GBs are typically thought to enhance ionic transport due to structural disorder and increased free volume. In this study, we report an unexpected anomaly: the $Σ3(111)$ GB in thorium dioxide (ThO$_2$), despite its compact and coherent structure, exhibits significantly higher oxygen ionic conductivity compared to the more open GB ($Σ19$ as an example). Using atomistic s…

3D PixBrush: Image-Guided Local Texture Synthesis
Dale Decatur, Itai Lang, Kfir Aberman, Rana Hanocka
https://arxiv.org/abs/2507.03731 https://

3D PixBrush: Image-Guided Local Texture Synthesis
We present 3D PixBrush, a method for performing image-driven edits of local regions on 3D meshes. 3D PixBrush predicts a localization mask and a synthesized texture that faithfully portray the object in the reference image. Our predicted localizations are both globally coherent and locally precise. Globally - our method contextualizes the object in the reference image and automatically positions it onto the input mesh. Locally - our method produces masks that conform to the geometry of the refe…

I didn't want to let the day go without posting a photo ;-) So I just grabbed one from last year when we visited #vienna in autumn.
A bit of context can be found on the blog: https://www.

Think Twice Before You Judge: Mixture of Dual Reasoning Experts for Multimodal Sarcasm Detection
Soumyadeep Jana, Abhrajyoti Kundu, Sanasam Ranbir Singh
https://arxiv.org/abs/2507.04458

Think Twice Before You Judge: Mixture of Dual Reasoning Experts for Multimodal Sarcasm Detection
Multimodal sarcasm detection has attracted growing interest due to the rise of multimedia posts on social media. Understanding sarcastic image-text posts often requires external contextual knowledge, such as cultural references or commonsense reasoning. However, existing models struggle to capture the deeper rationale behind sarcasm, relying mainly on shallow cues like image captions or object-attribute pairs from images. To address this, we propose \textbf{MiDRE} (\textbf{Mi}xture of \textbf{D…

10:42—Good Morning #Fediverse!
Being #ActuallyAutistic, I am currently hyperfixated on this image. I have been staring at it for an extended period, analyzing every detail.
I keep noticing new things every time I look at it. My brain is in “info-dump mode” and I could pr…

CP-Dilatation: A Copy-and-Paste Augmentation Method for Preserving the Boundary Context Information of Histopathology Images
Sungrae Hong, Sol Lee, Mun Yong Yi
https://arxiv.org/abs/2507.04660

CP-Dilatation: A Copy-and-Paste Augmentation Method for Preserving the Boundary Context Information of Histopathology Images
Medical AI diagnosis including histopathology segmentation has derived benefits from the recent development of deep learning technology. However, deep learning itself requires a large amount of training data and the medical image segmentation masking, in particular, requires an extremely high cost due to the shortage of medical specialists. To mitigate this issue, we propose a new data augmentation method built upon the conventional Copy and Paste (CP) augmentation technique, called CP-Dilatati…

A nonsmooth extension of the Brezzi-Rappaz-Raviart approximation theorem via metric regularity techniques and applications to nonlinear PDEs
Jules Berry (IRMAR, INSA Rennes, UR), Olivier Ley (INSA Rennes, IRMAR), Francisco Jos\'e Silva (XLIM, UNILIM)
https://arxiv.org/abs/2507.05774 …

A nonsmooth extension of the Brezzi-Rappaz-Raviart approximation theorem via metric regularity techniques and applications to nonlinear PDEs
We generalize the Brezzi-Rappaz-Raviart approximation theorem, which allows to obtain existence and a priori error estimates for approximations of solutions to some nonlinear partial differential equations. Our contribution lies in the fact that we typically allow for nonlinearities having merely Lipschitz regularity, while previous results required some form of differentiability. This is achieved by making use of the theory of metrically regular mappings, developed in the context of variationa…

Cross-Domain Image Synthesis: Generating H&E from Multiplex Biomarker Imaging
Jillur Rahman Saurav, Mohammad Sadegh Nasr, Jacob M. Luber
https://arxiv.org/abs/2508.04734 htt…

Cross-Domain Image Synthesis: Generating H&E from Multiplex Biomarker Imaging
While multiplex immunofluorescence (mIF) imaging provides deep, spatially-resolved molecular data, integrating this information with the morphological standard of Hematoxylin & Eosin (H&E) can be very important for obtaining complementary information about the underlying tissue. Generating a virtual H&E stain from mIF data offers a powerful solution, providing immediate morphological context. Crucially, this approach enables the application of the vast ecosystem of H&E-based computer-aided diag…

Unveiling the Potential of Diffusion Large Language Model in Controllable Generation
Zhen Xiong, Yujun Cai, Zhecheng Li, Yiwei Wang
https://arxiv.org/abs/2507.04504

Unveiling the Potential of Diffusion Large Language Model in Controllable Generation
Diffusion models, originally developed for image generation, have emerged as a promising alternative to autoregressive large language models (LLMs). We present a theoretical analysis comparing autoregressive and masked diffusion LLMs, revealing that the intrinsic bidirectional attention mechanism of diffusion LLMs (dLLMs) enables superior context modeling and generation controllability. However, existing dLLM applications face significant challenges in controllable generation: the native multi-…