Woojung Han

Woojung Han, Seil Kang, Youngjun Jun et al.

Image-to-Video diffusion models leverage input images to generate visually stunning content, yet frequently produce motion that violates physical laws. We reveal a surprising finding: a 2-step generation often exhibits better physical consistency than a 50-step output from the same model. Through spectral analysis, we trace this to phase erosion during denoising; the phase degrades significantly (dropping by $\approx 18\%$ from step 2 to step 50), whereas the magnitude remains relatively stable. Building on this insight, we propose PhaseLock, a training-free framework that preserves the valid motion priors from few-step inference throughout the denoising trajectory. Rather than relying on full-step inference for physical consistency, PhaseLock extracts a motion prior from just 2 steps and enforces it onto high-fidelity generation via Latent Delta Guidance. Our approach effectively mitigates phase degradation, improving physical consistency by an average of 6.2 points across diverse models while largely maintaining visual fidelity, with negligible overhead ($1.06\times$ time, $1.02\times$ memory) and reduced reliance on expensive external guidance methods ($\sim5\times$ time).

Seil Kang, Woojung Han, Junhyeok Kim et al.

We present an amortized framework for real-time visual attribution streaming in multimodal thinking models. When these models generate code from a screenshot or solve math problems from images, their long reasoning traces should be grounded in visual evidence. However, verifying this reliance is challenging: faithful causal methods require costly repeated backward passes or perturbations, while raw attention maps offer instant access, they lack causal validity. To resolve this, we introduce an amortized approach that learns to estimate the causal effects of semantic regions directly from the rich signals encoded in attention features. Across five diverse benchmarks and four thinking models, our approach achieves faithfulness comparable to exhaustive causal methods while enabling visual attribution streaming, where users observe grounding evidence as the model reasons, not after. Our results demonstrate that real-time, faithful attribution in multimodal thinking models is achievable through lightweight learning, not brute-force computation.

Youngjun Jun, Seil Kang, Woojung Han et al.

Video Diffusion Transformers (DiTs) have been synthesizing high-quality video with high fidelity from given text descriptions involving motion. However, understanding how Video DiTs convert motion words into video remains insufficient. Furthermore, while prior studies on interpretable saliency maps primarily target objects, motion-related behavior in Video DiTs remains largely unexplored. In this paper, we investigate concrete motion features that specify when and which object moves for a given motion concept. First, to spatially localize, we introduce GramCol, which adaptively produces per-frame saliency maps for any text concept, including both motion and non-motion. Second, we propose a motion-feature selection algorithm to obtain an Interpretable Motion-Attentive Map (IMAP) that localizes motion spatially and temporally. Our method discovers concept saliency maps without the need for any gradient calculation or parameter update. Experimentally, our method shows outstanding localization capability on the motion localization task and zero-shot video semantic segmentation, providing interpretable and clearer saliency maps for both motion and non-motion concepts.

Yeonkyung Lee, Woojung Han, Youngjun Jun et al.

Retinal foundation models have significantly advanced retinal image analysis by leveraging self-supervised learning to reduce dependence on labeled data while achieving strong generalization. Many recent approaches enhance retinal image understanding using report supervision, but obtaining clinical reports is often costly and challenging. In contrast, metadata (e.g., age, gender) is widely available and serves as a valuable resource for analyzing disease progression. To effectively incorporate patient-specific information, we propose PRETI, a retinal foundation model that integrates metadata-aware learning with robust self-supervised representation learning. We introduce Learnable Metadata Embedding (LME), which dynamically refines metadata representations. Additionally, we construct patient-level data pairs, associating images from the same individual to improve robustness against non-clinical variations. To further optimize retinal image representation, we propose Retina-Aware Adaptive Masking (RAAM), a strategy that selectively applies masking within the retinal region and dynamically adjusts the masking ratio during training. PRETI captures both global structures and fine-grained pathological details, resulting in superior diagnostic performance. Extensive experiments demonstrate that PRETI achieves state-of-the-art results across diverse diseases and biomarker predictions using in-house and public data, indicating the importance of metadata-guided foundation models in retinal disease analysis. Our code and pretrained model are available at https://github.com/MICV-yonsei/PRETI

Chanyoung Kim, Woojung Han, Dayun Ju et al.

Semantic segmentation has innately relied on extensive pixel-level annotated data, leading to the emergence of unsupervised methodologies. Among them, leveraging self-supervised Vision Transformers for unsupervised semantic segmentation (USS) has been making steady progress with expressive deep features. Yet, for semantically segmenting images with complex objects, a predominant challenge remains: the lack of explicit object-level semantic encoding in patch-level features. This technical limitation often leads to inadequate segmentation of complex objects with diverse structures. To address this gap, we present a novel approach, EAGLE, which emphasizes object-centric representation learning for unsupervised semantic segmentation. Specifically, we introduce EiCue, a spectral technique providing semantic and structural cues through an eigenbasis derived from the semantic similarity matrix of deep image features and color affinity from an image. Further, by incorporating our object-centric contrastive loss with EiCue, we guide our model to learn object-level representations with intra- and inter-image object-feature consistency, thereby enhancing semantic accuracy. Extensive experiments on COCO-Stuff, Cityscapes, and Potsdam-3 datasets demonstrate the state-of-the-art USS results of EAGLE with accurate and consistent semantic segmentation across complex scenes.

Woojung Han, Chanyoung Kim, Dayun Ju et al.

Recent advances in text-conditioned image generation diffusion models have begun paving the way for new opportunities in modern medical domain, in particular, generating Chest X-rays (CXRs) from diagnostic reports. Nonetheless, to further drive the diffusion models to generate CXRs that faithfully reflect the complexity and diversity of real data, it has become evident that a nontrivial learning approach is needed. In light of this, we propose CXRL, a framework motivated by the potential of reinforcement learning (RL). Specifically, we integrate a policy gradient RL approach with well-designed multiple distinctive CXR-domain specific reward models. This approach guides the diffusion denoising trajectory, achieving precise CXR posture and pathological details. Here, considering the complex medical image environment, we present "RL with Comparative Feedback" (RLCF) for the reward mechanism, a human-like comparative evaluation that is known to be more effective and reliable in complex scenarios compared to direct evaluation. Our CXRL framework includes jointly optimizing learnable adaptive condition embeddings (ACE) and the image generator, enabling the model to produce more accurate and higher perceptual CXR quality. Our extensive evaluation of the MIMIC-CXR-JPG dataset demonstrates the effectiveness of our RL-based tuning approach. Consequently, our CXRL generates pathologically realistic CXRs, establishing a new standard for generating CXRs with high fidelity to real-world clinical scenarios.

Chanyoung Kim, Dayun Ju, Woojung Han et al.

Open-Vocabulary Semantic Segmentation (OVSS) has advanced with recent vision-language models (VLMs), enabling segmentation beyond predefined categories through various learning schemes. Notably, training-free methods offer scalable, easily deployable solutions for handling unseen data, a key goal of OVSS. Yet, a critical issue persists: lack of object-level context consideration when segmenting complex objects in the challenging environment of OVSS based on arbitrary query prompts. This oversight limits models' ability to group semantically consistent elements within object and map them precisely to user-defined arbitrary classes. In this work, we introduce a novel approach that overcomes this limitation by incorporating object-level contextual knowledge within images. Specifically, our model enhances intra-object consistency by distilling spectral-driven features from vision foundation models into the attention mechanism of the visual encoder, enabling semantically coherent components to form a single object mask. Additionally, we refine the text embeddings with zero-shot object presence likelihood to ensure accurate alignment with the specific objects represented in the images. By leveraging object-level contextual knowledge, our proposed approach achieves state-of-the-art performance with strong generalizability across diverse datasets.

Woojung Han, Yeonkyung Lee, Chanyoung Kim et al.

Diffusion-based text-to-image (T2I) models have recently excelled in high-quality image generation, particularly in a training-free manner, enabling cost-effective adaptability and generalization across diverse tasks. However, while the existing methods have been continuously focusing on several challenges, such as "missing objects" and "mismatched attributes," another critical issue of "mislocated objects" remains where generated spatial positions fail to align with text prompts. Surprisingly, ensuring such seemingly basic functionality remains challenging in popular T2I models due to the inherent difficulty of imposing explicit spatial guidance via text forms. To address this, we propose STORM (Spatial Transport Optimization by Repositioning Attention Map), a novel training-free approach for spatially coherent T2I synthesis. STORM employs Spatial Transport Optimization (STO), rooted in optimal transport theory, to dynamically adjust object attention maps for precise spatial adherence, supported by a Spatial Transport (ST) Cost function that enhances spatial understanding. Our analysis shows that integrating spatial awareness is most effective in the early denoising stages, while later phases refine details. Extensive experiments demonstrate that STORM surpasses existing methods, effectively mitigating mislocated objects while improving missing and mismatched attributes, setting a new benchmark for spatial alignment in T2I synthesis.

Woojung Han, Seil Kang, Kyobin Choo et al.

Leveraging semantically precise pseudo masks derived from image-level class knowledge for segmentation, namely image-level Weakly Supervised Semantic Segmentation (WSSS), still remains challenging. While Class Activation Maps (CAMs) using CNNs have steadily been contributing to the success of WSSS, the resulting activation maps often narrowly focus on class-specific parts (e.g., only face of human). On the other hand, recent works based on vision transformers (ViT) have shown promising results based on their self-attention mechanism to capture the semantic parts but fail in capturing complete class-specific details (e.g., entire body parts of human but also with a dog nearby). In this work, we propose Complementary Branch (CoBra), a novel dual branch framework consisting of two distinct architectures which provide valuable complementary knowledge of class (from CNN) and semantic (from ViT) to each branch. In particular, we learn Class-Aware Projection (CAP) for the CNN branch and Semantic-Aware Projection (SAP) for the ViT branch to explicitly fuse their complementary knowledge and facilitate a new type of extra patch-level supervision. Our model, through CoBra, fuses CNN and ViT's complementary outputs to create robust pseudo masks that integrate both class and semantic information effectively. Extensive experiments qualitatively and quantitatively investigate how CNN and ViT complement each other on the PASCAL VOC 2012 dataset, showing a state-of-the-art WSSS result. This includes not only the masks generated by our model, but also the segmentation results derived from utilizing these masks as pseudo labels.

Seil Kang, Woojung Han, Dayun Ju et al.

Starting from flow- and diffusion-based transformers, Multi-modal Diffusion Transformers (MM-DiTs) have reshaped text-to-vision generation, gaining acclaim for exceptional visual fidelity. As these models advance, users continually push the boundary with imaginative or rare prompts, which advanced models still falter in generating, since their concepts are often too scarce to leave a strong imprint during pre-training. In this paper, we propose a simple yet effective intervention that surfaces rare semantics inside MM-DiTs without additional training steps, data, denoising-time optimization, or reliance on external modules (e.g., large language models). In particular, the joint-attention mechanism intrinsic to MM-DiT sequentially updates text embeddings alongside image embeddings throughout transformer blocks. We find that by mathematically expanding representational basins around text token embeddings via variance scale-up before the joint-attention blocks, rare semantics clearly emerge in MM-DiT's outputs. Furthermore, our results generalize effectively across text-to-vision tasks, including text-to-image, text-to-video, and text-driven image editing. Our work invites generative models to reveal the semantics that users intend, once hidden yet ready to surface.

Chanyoung Kim, Dayun Ju, Jinyeong Kim et al.

As recent text-conditioned diffusion models have enabled the generation of high-quality images, concerns over their potential misuse have also grown. This issue is critical in the medical domain, where text-conditioned generated medical images could enable insurance fraud or falsified records, highlighting the urgent need for reliable safeguards against unethical use. While watermarking techniques have emerged as a promising solution in general image domains, their direct application to medical imaging presents significant challenges. A key challenge is preserving fine-grained disease manifestations, as even minor distortions from a watermark may lead to clinical misinterpretation, which compromises diagnostic integrity. To overcome this gap, we present MedSign, a deep learning-based watermarking framework specifically designed for text-to-medical image synthesis, which preserves pathologically significant regions by adaptively adjusting watermark strength. Specifically, we generate a pathology localization map using cross-attention between medical text tokens and the diffusion denoising network, aggregating token-wise attention across layers, heads, and time steps. Leveraging this map, we optimize the LDM decoder to incorporate watermarking during image synthesis, ensuring cohesive integration while minimizing interference in diagnostically critical regions. Experimental results show that our MedSign preserves diagnostic integrity while ensuring watermark robustness, achieving state-of-the-art performance in image quality and detection accuracy on MIMIC-CXR and OIA-ODIR datasets.