arXiv subjects

Jianfu Zhang

Publications and source records attributed to Jianfu Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2025

Towards Explainable Fake Image Detection with Multi-Modal Large Language Models

Progress in image generation raises significant public security concerns. We argue that fake image detection should not operate as a "black box". Instead, an ideal approach must ensure both strong generalization and transparency. Recent progress in Multi-modal Large Language Models (MLLMs) offers new opportunities for reasoning-based AI-generated image detection. In this work, we evaluate the capabilities of MLLMs in comparison to traditional detection methods and human evaluators, highlighting their strengths and limitations. Furthermore, we design six distinct prompts and propose a framework that integrates these prompts to develop a more robust, explainable, and reasoning-driven detection system. The code is available at https://github.com/Gennadiyev/mllm-defake.

cs.CV↗

arXiv2025

DS-VTON: An Enhanced Dual-Scale Coarse-to-Fine Framework for Virtual Try-On

Despite recent progress, most existing virtual try-on methods still struggle to simultaneously address two core challenges: accurately aligning the garment image with the target human body, and preserving fine-grained garment textures and patterns. These two requirements map directly onto a coarse-to-fine generation paradigm, where the coarse stage handles structural alignment and the fine stage recovers rich garment details. Motivated by this observation, we propose DS-VTON, an enhanced dual-scale coarse-to-fine framework that tackles the try-on problem more effectively. DS-VTON consists of two stages: the first stage generates a low-resolution try-on result to capture the semantic correspondence between garment and body, where reduced detail facilitates robust structural alignment. In the second stage, a blend-refine diffusion process reconstructs high-resolution outputs by refining the residual between scales through noise-image blending, emphasizing texture fidelity and effectively correcting fine-detail errors from the low-resolution stage. In addition, our method adopts a fully mask-free generation strategy, eliminating reliance on human parsing maps or segmentation masks. Extensive experiments show that DS-VTON not only achieves state-of-the-art performance but consistently and significantly surpasses prior methods in both structural alignment and texture fidelity across multiple standard virtual try-on benchmarks.

cs.CV↗

arXiv2025

Influence of Density Distribution on Synchrotron Polarization Dispersion in Magnetized Interstellar Medium

Faraday rotation measure (RM) synthesis is a well-known approach originated in Burn (1966) and later developed by Brentjens \& de Bruyn (2005) for studying magnetic fields. This work presents a complementary approach--the polarization frequency analysis (PFA)--allowing for the properties of the turbulent magnetic field, which are difficult to include in Burn's original approach. Based on synthetic polarization observation of magnetohydrodynamic turbulence simulation data, we study the influence of the coupling effect between density and magnetic field on synchrotron polarization dispersion. By applying the PFA to different simulated interstellar turbulence environments, we find that the PFA technique can reveal the scaling slope of the turbulent magnetic field in the case of a weak coupling effect and can also reflect the scaling slope of the RM in the case of a strong coupling effect. Since it avoids the influence of Faraday depolarization, the PFA technique is a promising way to uncover turbulence properties using observational data from the Low-Frequency Array for Radio Astronomy and the Square Kilometre Array.

astro-ph.HE↗

arXiv2025

InterAnimate: Taming Region-aware Diffusion Model for Realistic Human Interaction Animation

Recent video generation research has focused heavily on isolated actions, leaving interactive motions-such as hand-face interactions-largely unexamined. These interactions are essential for emerging biometric authentication systems, which rely on interactive motion-based anti-spoofing approaches. From a security perspective, there is a growing need for large-scale, high-quality interactive videos to train and strengthen authentication models. In this work, we introduce a novel paradigm for animating realistic hand-face interactions. Our approach simultaneously learns spatio-temporal contact dynamics and biomechanically plausible deformation effects, enabling natural interactions where hand movements induce anatomically accurate facial deformations while maintaining collision-free contact. To facilitate this research, we present InterHF, a large-scale hand-face interaction dataset featuring 18 interaction patterns and 90,000 annotated videos. Additionally, we propose InterAnimate, a region-aware diffusion model designed specifically for interaction animation. InterAnimate leverages learnable spatial and temporal latents to effectively capture dynamic interaction priors and integrates a region-aware interaction mechanism that injects these priors into the denoising process. To the best of our knowledge, this work represents the first large-scale effort to systematically study human hand-face interactions. Qualitative and quantitative results show InterAnimate produces highly realistic animations, setting a new benchmark. Code and data will be made public to advance research.

cs.CV↗

arXiv2025

FW-VTON: Flattening-and-Warping for Person-to-Person Virtual Try-on

Traditional virtual try-on methods primarily focus on the garment-to-person try-on task, which requires flat garment representations. In contrast, this paper introduces a novel approach to the person-to-person try-on task. Unlike the garment-to-person try-on task, the person-to-person task only involves two input images: one depicting the target person and the other showing the garment worn by a different individual. The goal is to generate a realistic combination of the target person with the desired garment. To this end, we propose Flattening-and-Warping Virtual Try-On (\textbf{FW-VTON}), a method that operates in three stages: (1) extracting the flattened garment image from the source image; (2) warping the garment to align with the target pose; and (3) integrating the warped garment seamlessly onto the target person. To overcome the challenges posed by the lack of high-quality datasets for this task, we introduce a new dataset specifically designed for person-to-person try-on scenarios. Experimental evaluations demonstrate that FW-VTON achieves state-of-the-art performance, with superior results in both qualitative and quantitative assessments, and also excels in garment extraction subtasks.

cs.CV↗

arXiv2025

Studying the diffusion mechanism of cosmic-ray particles

More and more observations have indicated the existence of slow diffusion phenomena in astrophysical environments, such as around the supernova remnants and pulsar $γ$-ray halos, where the diffusion coefficient of cosmic rays (CRs) near the source region is significantly smaller than that far away from the source region. The inhomogeneous diffusion indicates the existence of multiple diffusion mechanisms.Comparing the CR mirror diffusion with the scattering one, we aim to explore their diffusion characteristics in different magnetohydrodynamic (MHD) turbulence regimes and understand the effect of different MHD modes on mirror and scattering diffusion. We perform numerical simulations with the test particle method. Within the global frame of reference, we first measure parallel and perpendicular CR diffusion and then determine the mean free path of CRs with varying energies.Our main results demonstrate that: (1) CRs experience a transition from superdiffusion to normal diffusion; (2) mirror diffusion is more important than scattering diffusion in confining CRs; (3) CR diffusion strongly depends on the properties of MHD turbulence; and (4) magnetosonic and Alfvén modes dominate the parallel and perpendicular diffusion of CR particles, respectively. The diffusion of CRs is a complex problem of mixing the mirror diffusion and scattering diffusion. The property of turbulent magnetic fields influences CR diffusion. The CR slow diffusion due to the presence of magnetic mirrors in turbulence has important implications for explaining observations near a CR source.

astro-ph.HE↗

arXiv2025

Interpretable and Reliable Detection of AI-Generated Images via Grounded Reasoning in MLLMs

The rapid advancement of image generation technologies intensifies the demand for interpretable and robust detection methods. Although existing approaches often attain high accuracy, they typically operate as black boxes without providing human-understandable justifications. Multi-modal Large Language Models (MLLMs), while not originally intended for forgery detection, exhibit strong analytical and reasoning capabilities. When properly fine-tuned, they can effectively identify AI-generated images and offer meaningful explanations. However, existing MLLMs still struggle with hallucination and often fail to align their visual interpretations with actual image content and human reasoning. To bridge this gap, we construct a dataset of AI-generated images annotated with bounding boxes and descriptive captions that highlight synthesis artifacts, establishing a foundation for human-aligned visual-textual grounded reasoning. We then finetune MLLMs through a multi-stage optimization strategy that progressively balances the objectives of accurate detection, visual localization, and coherent textual explanation. The resulting model achieves superior performance in both detecting AI-generated images and localizing visual flaws, significantly outperforming baseline methods.

cs.CV↗

arXiv2025

Insight into the origin of multiwavelength emissions of PKS 1510-089 through modeling 12 SEDs from 2008 to 2015

PKS\,1510$-$089 is one of the most peculiar sources among the FSRQs, exhibiting a notable big blue bump (BBB). This provides an unique opportunity to explore the coupling between the activity of the central engine and the relativistic jet, offering further insight into the origin of the multiwavelength emissions. To this end, we collected multiwavelength data spanning four periods from 2008 to 2015 and performed the spectral energy distribution (SED) modeling using a one-zone homogeneous leptonic model. In the model, a multichromatic accretion disk (AD) is used to fit the optical/UV data sets, while the external radiation fields from the broad-line region (BLR) and dusty torus (DT) are properly considered to produce the high-energy $γ$-ray emissions. Our best fit to 12 SEDs yields the following results: (i) The innermost stable orbit ($R_{\rm ISO}$) of the AD is not stable but varies between $3\,R_{\rm S}$ and $18\,R_{\rm S}$ during these observations. (ii) The high-energy hump of the SED is well dominated by Compton scattering of the BLR photons, while the X-ray flux may be comprised of multiple radiation components. (iii) The $γ$-ray emitting regions are generally matter-dominated, with low magnetization, and are located beyond the BLR but within the DT. At such distance, the multiwavelength emissions are likely to originate from shock accelerations; (iv) For the energization of the relativistic jet, our study supports the Blandford$-$Znajek (BZ) mechanism, instead of the Blandford$-$Payne (BP) mechanism, as the latter fails to power the jet.

astro-ph.HE↗

arXiv2025

Revealing the spectral properties of magnetic turbulence by synchrotron polarization gradients

Based on the modern understanding of MHD turbulence theory, we propose a new method for measuring the spectral properties of magnetic turbulence by synchrotron polarization gradient analysis. Using synthetic polarization observational data, we first confirm the feasibility of the gradient technique to determine the scaling properties of magnetic turbulence. We then apply this technique to the Galactic plane survey data from the Australia Telescope Compact Array (ATCA) and obtain the power-law spectral feature of the Galactic magnetic turbulence of $E\propto k^{-10/3}$. Gradient techniques open up a new way to obtain spectral properties of magnetic turbulence from polarization observations, such as the Low-Frequency Array for Radio Astronomy (LOFAR), in the presence of low-frequency and strong Faraday depolarization.

astro-ph.HE↗

arXiv2024

DomainGallery: Few-shot Domain-driven Image Generation by Attribute-centric Finetuning

The recent progress in text-to-image models pretrained on large-scale datasets has enabled us to generate various images as long as we provide a text prompt describing what we want. Nevertheless, the availability of these models is still limited when we expect to generate images that fall into a specific domain either hard to describe or just unseen to the models. In this work, we propose DomainGallery, a few-shot domain-driven image generation method which aims at finetuning pretrained Stable Diffusion on few-shot target datasets in an attribute-centric manner. Specifically, DomainGallery features prior attribute erasure, attribute disentanglement, regularization and enhancement. These techniques are tailored to few-shot domain-driven generation in order to solve key issues that previous works have failed to settle. Extensive experiments are given to validate the superior performance of DomainGallery on a variety of domain-driven generation scenarios. Codes are available at https://github.com/Ldhlwh/DomainGallery.

cs.CV↗

arXiv2024

DeMamba: AI-Generated Video Detection on Million-Scale GenVideo Benchmark

Recently, video generation techniques have advanced rapidly. Given the popularity of video content on social media platforms, these models intensify concerns about the spread of fake information. Therefore, there is a growing demand for detectors capable of distinguishing between fake AI-generated videos and mitigating the potential harm caused by fake information. However, the lack of large-scale datasets from the most advanced video generators poses a barrier to the development of such detectors. To address this gap, we introduce the first AI-generated video detection dataset, GenVideo. It features the following characteristics: (1) a large volume of videos, including over one million AI-generated and real videos collected; (2) a rich diversity of generated content and methodologies, covering a broad spectrum of video categories and generation techniques. We conducted extensive studies of the dataset and proposed two evaluation methods tailored for real-world-like scenarios to assess the detectors' performance: the cross-generator video classification task assesses the generalizability of trained detectors on generators; the degraded video classification task evaluates the robustness of detectors to handle videos that have degraded in quality during dissemination. Moreover, we introduced a plug-and-play module, named Detail Mamba (DeMamba), designed to enhance the detectors by identifying AI-generated videos through the analysis of inconsistencies in temporal and spatial dimensions. Our extensive experiments demonstrate DeMamba's superior generalizability and robustness on GenVideo compared to existing detectors. We believe that the GenVideo dataset and the DeMamba module will significantly advance the field of AI-generated video detection. Our code and dataset will be aviliable at \url{https://github.com/chenhaoxing/DeMamba}.

cs.CV↗

arXiv2024

Self-Supervised Vision Transformer for Enhanced Virtual Clothes Try-On

Virtual clothes try-on has emerged as a vital feature in online shopping, offering consumers a critical tool to visualize how clothing fits. In our research, we introduce an innovative approach for virtual clothes try-on, utilizing a self-supervised Vision Transformer (ViT) coupled with a diffusion model. Our method emphasizes detail enhancement by contrasting local clothing image embeddings, generated by ViT, with their global counterparts. Techniques such as conditional guidance and focus on key regions have been integrated into our approach. These combined strategies empower the diffusion model to reproduce clothing details with increased clarity and realism. The experimental results showcase substantial advancements in the realism and precision of details in virtual try-on experiences, significantly surpassing the capabilities of existing technologies.

cs.CV↗

arXiv2024

Supervised Contrastive Learning for Snapshot Spectral Imaging Face Anti-Spoofing

This study reveals a cutting-edge re-balanced contrastive learning strategy aimed at strengthening face anti-spoofing capabilities within facial recognition systems, with a focus on countering the challenges posed by printed photos, and highly realistic silicone or latex masks. Leveraging the HySpeFAS dataset, which benefits from Snapshot Spectral Imaging technology to provide hyperspectral images, our approach harmonizes class-level contrastive learning with data resampling and an innovative real-face oriented reweighting technique. This method effectively mitigates dataset imbalances and reduces identity-related biases. Notably, our strategy achieved an unprecedented 0.0000\% Average Classification Error Rate (ACER) on the HySpeFAS dataset, ranking first at the Chalearn Snapshot Spectral Imaging Face Anti-spoofing Challenge on CVPR 2024.

cs.CV↗

arXiv2024

User-Friendly Customized Generation with Multi-Modal Prompts

Text-to-image generation models have seen considerable advancement, catering to the increasing interest in personalized image creation. Current customization techniques often necessitate users to provide multiple images (typically 3-5) for each customized object, along with the classification of these objects and descriptive textual prompts for scenes. This paper questions whether the process can be made more user-friendly and the customization more intricate. We propose a method where users need only provide images along with text for each customization topic, and necessitates only a single image per visual concept. We introduce the concept of a ``multi-modal prompt'', a novel integration of text and images tailored to each customization concept, which simplifies user interaction and facilitates precise customization of both objects and scenes. Our proposed paradigm for customized text-to-image generation surpasses existing finetune-based methods in user-friendliness and the ability to customize complex objects with user-friendly inputs. Our code is available at $\href{https://github.com/zhongzero/Multi-Modal-Prompt}{https://github.com/zhongzero/Multi-Modal-Prompt}$.

cs.CV↗

arXiv2024

Assessing Image Inpainting via Re-Inpainting Self-Consistency Evaluation

Image inpainting, the task of reconstructing missing segments in corrupted images using available data, faces challenges in ensuring consistency and fidelity, especially under information-scarce conditions. Traditional evaluation methods, heavily dependent on the existence of unmasked reference images, inherently favor certain inpainting outcomes, introducing biases. Addressing this issue, we introduce an innovative evaluation paradigm that utilizes a self-supervised metric based on multiple re-inpainting passes. This approach, diverging from conventional reliance on direct comparisons in pixel or feature space with original images, emphasizes the principle of self-consistency to enable the exploration of various viable inpainting solutions, effectively reducing biases. Our extensive experiments across numerous benchmarks validate the alignment of our evaluation method with human judgment.

cs.CV↗

arXiv2024

WildFake: A Large-scale Challenging Dataset for AI-Generated Images Detection

The extraordinary ability of generative models enabled the generation of images with such high quality that human beings cannot distinguish Artificial Intelligence (AI) generated images from real-life photographs. The development of generation techniques opened up new opportunities but concurrently introduced potential risks to privacy, authenticity, and security. Therefore, the task of detecting AI-generated imagery is of paramount importance to prevent illegal activities. To assess the generalizability and robustness of AI-generated image detection, we present a large-scale dataset, referred to as WildFake, comprising state-of-the-art generators, diverse object categories, and real-world applications. WildFake dataset has the following advantages: 1) Rich Content with Wild collection: WildFake collects fake images from the open-source community, enriching its diversity with a broad range of image classes and image styles. 2) Hierarchical structure: WildFake contains fake images synthesized by different types of generators from GANs, diffusion models, to other generative models. These key strengths enhance the generalization and robustness of detectors trained on WildFake, thereby demonstrating WildFake's considerable relevance and effectiveness for AI-generated detectors in real-world scenarios. Moreover, our extensive evaluation experiments are tailored to yield profound insights into the capabilities of different levels of generative models, a distinctive advantage afforded by WildFake's unique hierarchical structure.

cs.CV↗

arXiv2024

ComFusion: Personalized Subject Generation in Multiple Specific Scenes From Single Image

Recent advancements in personalizing text-to-image (T2I) diffusion models have shown the capability to generate images based on personalized visual concepts using a limited number of user-provided examples. However, these models often struggle with maintaining high visual fidelity, particularly in manipulating scenes as defined by textual inputs. Addressing this, we introduce ComFusion, a novel approach that leverages pretrained models generating composition of a few user-provided subject images and predefined-text scenes, effectively fusing visual-subject instances with textual-specific scenes, resulting in the generation of high-fidelity instances within diverse scenes. ComFusion integrates a class-scene prior preservation regularization, which leverages composites the subject class and scene-specific knowledge from pretrained models to enhance generation fidelity. Additionally, ComFusion uses coarse generated images, ensuring they align effectively with both the instance image and scene texts. Consequently, ComFusion maintains a delicate balance between capturing the essence of the subject and maintaining scene fidelity.Extensive evaluations of ComFusion against various baselines in T2I personalization have demonstrated its qualitative and quantitative superiority.

cs.CV↗

arXiv2023

Dataset Distillation in Latent Space

Dataset distillation (DD) is a newly emerging research area aiming at alleviating the heavy computational load in training models on large datasets. It tries to distill a large dataset into a small and condensed one so that models trained on the distilled dataset can perform comparably with those trained on the full dataset when performing downstream tasks. Among the previous works in this area, there are three key problems that hinder the performance and availability of the existing DD methods: high time complexity, high space complexity, and low info-compactness. In this work, we simultaneously attempt to settle these three problems by moving the DD processes from conventionally used pixel space to latent space. Encoded by a pretrained generic autoencoder, latent codes in the latent space are naturally info-compact representations of the original images in much smaller sizes. After transferring three mainstream DD algorithms to latent space, we significantly reduce time and space consumption while achieving similar performance, allowing us to distill high-resolution datasets or target at greater data ratio that previous methods have failed. Besides, within the same storage budget, we can also quantitatively deliver more latent codes than pixel-level images, which further boosts the performance of our methods.

cs.CV↗