arXiv subjects

Taewon Kang

Publications and source records attributed to Taewon Kang.

At least 19 recordsLinked to original sources

arXiv2026

TPD: Temporal Prior Decoupling for Text-to-Video Diffusion Models

Text-to-video diffusion models generate temporally coherent content from natural language, yet when a prompt describes an early scene that persists while a new event emerges on top of it---such as "a tall sandcastle standing on a beach where a wave rushes in and washes it away"---generation frequently fails to realize the late-segment event in the corresponding frames. We identify this failure as Temporal Prior Suppression (TPS): the dominant prior of the early segment captures the cross-attention trajectory across the temporal axis and suppresses the guidance signal needed for late-segment realization, a competing tendency existing guidance mechanisms do not model. We introduce Temporal Prior Decoupling (TPD), a training-free framework that restores suppressed late-segment signals during diffusion sampling. TPD constructs a temporal counterfactual by conditioning on the early segment alone, and defines the discrepancy between the full-prompt and counterfactual trajectories as a suppressed signal direction. Rather than removing this direction as in prior subtractive projection methods, TPD restores it through a frame-selective lower-bound constraint resolved jointly over diffusion timestep and video frame, realizing the suppressed event in the late frames without disrupting early-segment coherence: where prior work enforces upper-bound feasibility to remove unwanted semantics, TPD enforces lower-bound feasibility to guarantee suppressed-signal contribution. TPD runs entirely within standard diffusion sampling without retraining, and is defined purely in classifier-free guidance space, making it backbone-agnostic by construction. Experiments show that TPD significantly improves late-concept realization while preserving temporal coherence and visual fidelity, and that the targeted suppression recurs across distinct text-to-video backbones.

cs.CV↗

arXiv2026

DCR: Counterfactual Attractor Guidance for Rare Compositional Generation

Diffusion models generate realistic visual content, yet often fail to produce rare but plausible compositions. When prompted with combinations that are valid but underrepresented in training data, such as a snowy beach or a rainbow at night, the generation process frequently collapses toward more common alternatives. We identify this failure mode as default completion bias, where denoising trajectories are implicitly attracted toward high-frequency semantic configurations. Existing guidance mechanisms do not explicitly model this competing tendency and therefore struggle to prevent such collapse. We introduce Default Completion Repulsion (DCR), a training-free framework that explicitly models and suppresses default completion behavior. DCR constructs a counterfactual attractor by relaxing the rare compositional factor while preserving surrounding semantics, inducing an alternative denoising trajectory reflecting the model's preferred completion. We define the discrepancy between target and attractor trajectories as a counterfactual drift, and propose a projection-based repulsion mechanism that removes guidance components aligned with this drift direction. This suppresses undesired frequent completions while preserving other semantic components. DCR operates entirely within the standard diffusion sampling process without retraining or architectural modification. Experiments on rare compositional prompts show that DCR improves compositional fidelity while maintaining visual quality. Our analysis further shows that the framework exposes and counteracts intrinsic model biases, offering a new perspective on controllable generation beyond explicit constraint enforcement.

cs.CV↗

arXiv2026

NEGATE: Constrained Semantic Guidance for Linguistic Negation in Text-to-Video Diffusion

Negation is a fundamental linguistic operator, yet it remains inadequately modeled in diffusion-based generative systems. In this work, we present a formal treatment of linguistic negation in diffusion-based generative models by modeling it as a structured feasibility constraint on semantic guidance within diffusion dynamics. Rather than introducing heuristics or retraining model parameters, we reinterpret classifier-free guidance as defining a semantic update direction and enforce negation by projecting the update onto a convex constraint set derived from linguistic structure. This novel formulation provides a unified framework for handling diverse negation phenomena, including object absence, graded non-inversion semantics, multi-negation composition, and scope-sensitive disambiguation. Our approach is training-free, compatible with pretrained diffusion backbones, and naturally extends from image generation to temporally evolving video trajectories. In addition, we introduce a structured negation-centric benchmark suite that isolates distinct linguistic failure modes in generative systems, to further research in this area. Experiments demonstrate that our method achieves robust negation compliance while preserving visual fidelity and structural coherence, establishing the first unified formulation of linguistic negation in diffusion-based generative models beyond representation-level evaluation.

cs.CV↗

arXiv2026

Trajectory-Guided Diffusion for Foreground-Preserving Background Generation in Multi-Layer Documents

We present a diffusion-based framework for document-centric background generation that achieves foreground preservation and multi-page stylistic consistency through latent-space design rather than explicit constraints. Instead of suppressing diffusion updates or applying masking heuristics, our approach reinterprets diffusion as the evolution of stochastic trajectories through a structured latent space. By shaping the initial noise and its geometric alignment, background generation naturally avoids designated foreground regions, allowing readable content to remain intact without auxiliary mechanisms. To address the long-standing issue of stylistic drift across pages, we decouple style control from text conditioning and introduce cached style directions as persistent vectors in latent space. Once selected, these directions constrain diffusion trajectories to a shared stylistic subspace, ensuring consistent appearance across pages and editing iterations. This formulation eliminates the need for repeated prompt-based style specification and provides a more stable foundation for multi-page generation. Our framework admits a geometric and physical interpretation, where diffusion paths evolve on a latent manifold shaped by preferred directions, and foreground regions are rarely traversed as a consequence of trajectory initialization rather than explicit exclusion. The proposed method is training-free, compatible with existing diffusion backbones, and produces visually coherent, foreground-preserving results across complex documents. By reframing diffusion as trajectory design in latent space, we offer a principled approach to consistent and structured generative modeling.

cs.CV↗

arXiv2025

Text-Conditioned Background Generation for Editable Multi-Layer Documents

We present a framework for document-centric background generation with multi-page editing and thematic continuity. To ensure text regions remain readable, we employ a latent masking formulation that softly attenuates updates in the diffusion space, inspired by smooth barrier functions in physics and numerical optimization. In addition, we introduce Automated Readability Optimization (ARO), which automatically places semi-transparent, rounded backing shapes behind text regions. ARO determines the minimal opacity needed to satisfy perceptual contrast standards (WCAG 2.2) relative to the underlying background, ensuring readability while maintaining aesthetic harmony without human intervention. Multi-page consistency is maintained through a summarization-and-instruction process, where each page is distilled into a compact representation that recursively guides subsequent generations. This design reflects how humans build continuity by retaining prior context, ensuring that visual motifs evolve coherently across an entire document. Our method further treats a document as a structured composition in which text, figures, and backgrounds are preserved or regenerated as separate layers, allowing targeted background editing without compromising readability. Finally, user-provided prompts allow stylistic adjustments in color and texture, balancing automated consistency with flexible customization. Our training-free framework produces visually coherent, text-preserving, and thematically aligned documents, bridging generative modeling with natural design workflows.

cs.CV↗

arXiv2025

HART: Human Aligned Reconstruction Transformer

We introduce HART, a unified framework for sparse-view human reconstruction. Given a small set of uncalibrated RGB images of a person as input, it outputs a watertight clothed mesh, the aligned SMPL-X body mesh, and a Gaussian-splat representation for photorealistic novel-view rendering. Prior methods for clothed human reconstruction either optimize parametric templates, which overlook loose garments and human-object interactions, or train implicit functions under simplified camera assumptions, limiting applicability in real scenes. In contrast, HART predicts per-pixel 3D point maps, normals, and body correspondences, and employs an occlusion-aware Poisson reconstruction to recover complete geometry, even in self-occluded regions. These predictions also align with a parametric SMPL-X body model, ensuring that reconstructed geometry remains consistent with human structure while capturing loose clothing and interactions. These human-aligned meshes initialize Gaussian splats to further enable sparse-view rendering. While trained on only 2.3K synthetic scans, HART achieves state-of-the-art results: Chamfer Distance improves by 18-23 percent for clothed-mesh reconstruction, PA-V2V drops by 6-27 percent for SMPL-X estimation, LPIPS decreases by 15-27 percent for novel-view synthesis on a wide range of datasets. These results suggest that feed-forward transformers can serve as a scalable model for robust human reconstruction in real-world settings. Code and models will be released.

cs.CV↗

arXiv2025

Correlation Analysis Between MF R-Mode Temporal ASF and Meteorological Factors

As the vulnerabilities of global navigation satellite systems (GNSS) have become more widely recognized, the need for complementary navigation systems has grown. Medium frequency ranging mode (MF R-Mode) has gained attention as an effective backup system during GNSS outages, owing to its strong signal strength and cost-effective scalability. However, to achieve accurate positioning, MF R-Mode requires correction for the additional secondary factor (ASF), a propagation delay affected by terrain. The temporal variation of ASF, known as temporal ASF, is typically corrected using reference stations; however, the effectiveness of this method decreases with distance from the reference station. In this study, we analyzed the correlation between temporal ASF and meteorological factors to evaluate the feasibility of predicting temporal ASF based on meteorological factors. Among these factors, temperature and humidity showed significant correlations with temporal ASF, suggesting their potential utility in ASF correction.

eess.SP↗

arXiv2025

Mapping Innovation Networks: A Network-Based Approach to Actor Heterogeneity in National Innovation Systems

The Triple Helix model has provided a foundational framework for analyzing National Innovation Systems by highlighting the roles of universities, industries, and government research institutes. However, increasing heterogeneity within these actor groups limits the explanatory power of typological approaches. This study introduces a capability-based network methodology that maps the structural relationships among innovation actors based on the similarity of their research and development (R&D) capabilities. Drawing on Economic Complexity Theory, we measure each actor's revealed comparative advantage (RCA) across scientific and technological fields and construct an R&D Actor Space - a proximity-based network that reflects the relational configuration of innovation capacities. Applying this method to Korean R&D data, we uncover a stratified system in which central, highly diversified universities coexist with more specialized firms and government institutes. Network analysis reveals assortative and unequal structures, and hierarchical clustering further highlights layered subgroupings. By moving beyond categorical classification, this capability-based network approach provides a scalable and generalizable tool for analyzing structural complexity within national innovation systems.

physics.soc-ph↗

arXiv2025

Monocular Vision-Based Swarm Robot Localization Using Equilateral Triangular Formations

Localization of mobile robots is crucial for deploying robots in real-world applications such as search and rescue missions. This work aims to develop an accurate localization system applicable to swarm robots equipped only with low-cost monocular vision sensors and visual markers. The system is designed to operate in fully open spaces, without landmarks or support from positioning infrastructures. To achieve this, we propose a localization method based on equilateral triangular formations. By leveraging the geometric properties of equilateral triangles, the accurate two-dimensional position of each participating robot is estimated using one-dimensional lateral distance information between robots, which can be reliably and accurately obtained with a low-cost monocular vision sensor. Experimental and simulation results demonstrate that, as travel time increases, the positioning error of the proposed method becomes significantly smaller than that of a conventional dead-reckoning system, another low-cost localization approach applicable to open environments.

cs.RO↗

arXiv2025

Enhancing eLoran Timing Accuracy via Machine Learning with Meteorological and Terrain Data

The vulnerabilities of global navigation satellite systems (GNSS) to signal interference have increased the demand for complementary positioning, navigation, and timing (PNT) systems. To address this, South Korea has decided to deploy an enhanced long-range navigation (eLoran) system as a complementary PNT solution. Similar to GNSS, eLoran provides highly accurate timing information, which is essential for applications such as telecommunications, financial systems, and power distribution. However, the primary sources of error for GNSS and eLoran differ. For eLoran, the main source of error is signal propagation delay over land, known as the additional secondary factor (ASF). This delay, influenced by ground conductivity and weather conditions along the signal path, is challenging to predict and mitigate. In this paper, we measure the time difference (TD) between GPS and eLoran using a time interval counter and analyze the correlations between eLoran/GPS TD and eleven meteorological factors. Accurate estimation of eLoran/GPS TD could enable eLoran to achieve timing accuracy comparable to that of GPS. We propose two estimation models for eLoran/GPS TD and compare their performance with existing TD estimation methods. The proposed WLR-AGRNN model captures the linear relationships between meteorological factors and eLoran/GPS TD using weighted linear regression (WLR) and models nonlinear relationships between outputs from expert networks through an anisotropic general regression neural network (AGRNN). The model incorporates terrain elevation to appropriately weight meteorological data, as elevation influences signal propagation delay. Experimental results based on four months of data demonstrate that the WLR-AGRNN model outperforms other models, highlighting its effectiveness in improving eLoran/GPS TD estimation accuracy.

eess.SP↗

arXiv2025

Character-Centered Dialogue Generation from Scene-Level Prompts

Recent advances in scene-based video generation enable coherent visual narratives from structured prompts, yet a key aspect of storytelling -- character-driven dialogue and speech -- remains underexplored. We present a modular pipeline that transforms action-level prompts into visually and auditorily grounded dialogue, enriching scene-based storytelling with natural voice and character expression. Our method takes a pair of prompts per scene, defining the setting and character behavior. While a story generation model such as Text2Story produces the visual scene, we focus on generating expressive, character-consistent utterances grounded in both the prompts and a representative scene image. A pretrained vision-language encoder extracts high-level visual semantics, which are combined with structured prompts to guide a large language model for dialogue synthesis. To maintain contextual and emotional consistency across scenes, we introduce a Recursive Narrative Bank, a speaker-aware, temporally structured memory that accumulates each character's dialogue history. Inspired by Script Theory, this design enables dialogue that reflects evolving goals, social context, and narrative roles. Finally, we render each utterance as expressive, character-conditioned speech, producing fully voiced, multimodal video narratives. Our training-free framework generalizes across diverse story settings, providing a scalable solution for coherent, character-grounded audiovisual storytelling.

cs.CV↗

arXiv2025

Scene-Action Prompt Fusion for Coherent Text-to-Video Storytelling

Generating coherent long-form video sequences from discrete text prompts remains challenging due to difficulties in maintaining temporal coherence, semantic consistency, and scene-action continuity across segments. We propose a novel storytelling framework that integrates scene and action prompts through dynamics-inspired prompt mixing. Our approach combines three key components: (i) a bidirectional time-weighted latent blending strategy that enforces temporal consistency between consecutive video segments, (ii) a dynamics-informed prompt weighting (DIPW) mechanism that adaptively balances scene and action prompts at each diffusion timestep based on CLIP-based alignment, narrative progression, and temporal smoothness, and (iii) a semantic action representation that encodes high-level action semantics to modulate transitions according to action similarity. Latent-space blending preserves spatial coherence within scenes, while time-weighted blending introduces bidirectional temporal constraints to prevent abrupt transitions. Together, these components enable fluid and coherent video narratives that faithfully reflect both scene context and action dynamics. Extensive experiments demonstrate that our method significantly outperforms baselines, producing temporally consistent and visually compelling long-form videos without any additional training, thereby bridging the gap between short clips and extended text-driven video storytelling.

cs.CV↗

arXiv2024

3D-free meets 3D priors: Novel View Synthesis from a Single Image with Pretrained Diffusion Guidance

Recent 3D novel view synthesis (NVS) methods often require extensive 3D data for training, and also typically lack generalization beyond the training distribution. Moreover, they tend to be object centric and struggle with complex and intricate scenes. Conversely, 3D-free methods can generate text-controlled views of complex, in-the-wild scenes using a pretrained stable diffusion model without the need for a large amount of 3D-based training data, but lack camera control. In this paper, we introduce a method capable of generating camera-controlled viewpoints from a single input image, by combining the benefits of 3D-free and 3D-based approaches. Our method excels in handling complex and diverse scenes without extensive training or additional 3D and multiview data. It leverages widely available pretrained NVS models for weak guidance, integrating this knowledge into a 3D-free view synthesis style approach, along with enriching the CLIP vision-language space with 3D camera angle information, to achieve the desired results. Experimental results demonstrate that our method outperforms existing models in both qualitative and quantitative evaluations, achieving high-fidelity, consistent novel view synthesis at desired camera angles across a wide variety of scenes while maintaining accurate, natural detail representation and image clarity across various viewpoints. We also support our method with a comprehensive analysis of 2D image generation models and the 3D space, providing a solid foundation and rationale for our solution.

cs.CV↗

arXiv2024

Structure-based Optical Logics Without Using Transistors

The commercialization of transistors capable of both switching and amplification in 1960 resulted in the development of second-generation computers, which resulted in the miniaturization and lightening while accelerating the reduction and development of production costs. However, the self-resistance and the resistance used in conjunction with semiconductors, which are the basic principles of computers, generate a lot of heat, which results in semiconductor obsolescence, and limits the computation speed (clock rate). In implementing logic operation, this paper proposes the concept of Structure-based Computer which can implement NOT gate made of semiconductor transistor only by Structure-based twist of cable without resistance. In Structure-based computer, the theory of 'inverse signal pair' of digital signals was introduced so that it could operate in a different way than semiconductor-based transistors. In this paper, we propose a new hardware called Structure-based computer that can solve various problems in semiconductor computers only with the wiring structure of the conductor itself, not with the silicon-based semiconductor. Furthermore, we propose a deep-priority exploration-based simulation method that can easily implement and test complex Structure-based computer circuits. Furthermore, this paper suggests a mechanism to implement optical computers currently under development and research based on structures rather than devices.

cs.ET↗

arXiv2023

Detection of Pedestrian Turning Motions to Enhance Indoor Map Matching Performance

A pedestrian navigation system (PNS) in indoor environments, where global navigation satellite system (GNSS) signal access is difficult, is necessary, particularly for search and rescue (SAR) operations in large buildings. This paper focuses on studying pedestrian walking behaviors to enhance the performance of indoor pedestrian dead reckoning (PDR) and map matching techniques. Specifically, our research aims to detect pedestrian turning motions using smartphone inertial measurement unit (IMU) information in a given PDR trajectory. To improve existing methods, including the threshold-based turn detection method, hidden Markov model (HMM)-based turn detection method, and pruned exact linear time (PELT) algorithm-based turn detection method, we propose enhanced algorithms that better detect pedestrian turning motions. During field tests, using the threshold-based method, we observed a missed detection rate of 20.35% and a false alarm rate of 7.65%. The PELT-based method achieved a significant improvement with a missed detection rate of 8.93% and a false alarm rate of 6.97%. However, the best results were obtained using the HMM-based method, which demonstrated a missed detection rate of 5.14% and a false alarm rate of 2.00%. In summary, our research contributes to the development of a more accurate and reliable pedestrian navigation system by leveraging smartphone IMU data and advanced algorithms for turn detection in indoor environments.

eess.SP↗

arXiv2023

GeoT: A Geometry-aware Transformer for Reliable Molecular Property Prediction and Chemically Interpretable Representation Learning

In recent years, molecular representation learning has emerged as a key area of focus in various chemical tasks. However, many existing models fail to fully consider the geometric information of molecular structures, resulting in less intuitive representations. Moreover, the widely used message-passing mechanism is limited to provide the interpretation of experimental results from a chemical perspective. To address these challenges, we introduce a novel Transformer-based framework for molecular representation learning, named the Geometry-aware Transformer (GeoT). GeoT learns molecular graph structures through attention-based mechanisms specifically designed to offer reliable interpretability, as well as molecular property prediction. Consequently, GeoT can generate attention maps of interatomic relationships associated with training objectives. In addition, GeoT demonstrates comparable performance to MPNN-based models while achieving reduced computational complexity. Our comprehensive experiments, including an empirical simulation, reveal that GeoT effectively learns the chemical insights into molecular structures, bridging the gap between artificial intelligence and molecular sciences.

cs.LG↗

arXiv2022

Construction of the Radio Map with Defective GPS Position Information

The basic idea of RSS-based indoor positioning is to estimate the receiver location by matching the measured received signal strength indicator (RSSI) with preestablished RSSI collections with corresponding locations, known as the radio map. Hence, constructing an accurate radio map directly relates to accurate positioning performance in RSS-based indoor positioning. RSS-based indoor positioning can be easily conducted with a radio map that surveys every location, but a complete radio map cannot be constructed when the map area includes locations that are physically impossible to reach or denied access. In addition, measurement errors or device problems can occur during the survey, resulting in degradation of the radio map. We analyzed incidents that occurred in actual RSSI surveys that could disrupt the construction of the radio map and proposed methods to construct a more accurate radio map.

eess.SY↗

arXiv2022

Evaluation of RF Fingerprinting-Aided RSS-Based Target Localization for Emergency Response

Target localization is essential for emergency dispatching situations. Maximum likelihood estimation (MLE) methods are widely used to estimate the target position based on the received signal strength measurements. However, the performance of MLE solvers is significantly affected by the initialization (i.e., initial guess of the solution or solution search space). To address this, a previous study proposed the semidefinite programming (SDP)-based MLE initialization. However, the performance of the SDP-based initialization technique is largely affected by the shadowing variance and geometric diversity between the target and receivers. In this study, a radio frequency (RF) fingerprinting-based MLE initialization is proposed. Further, a maximum likelihood problem for target localization combining RF fingerprinting is formulated. In the three test environments of open space, urban, and indoor, the proposed RF fingerprinting-aided target localization method showed a performance improvement of up to 63.31% and an average of 39.13%, compared to the MLE algorithm initialized with SDP. Furthermore, unlike the SDP-MLE method, the proposed method was not significantly affected by the poor geometry between the target and receivers in our experiments.

eess.SP↗