Searcharxiv⌕ Search

arXiv subjects

Sang-Il Choi

Publications and source records attributed to Sang-Il Choi.

6 recordsLinked to original sources

Nunchi: Ears Help Eyes in Multi-Party Conversations

Identifying whom a person looks at is central to understanding multi-party conversations, yet gaze geometry alone becomes ambiguous as more people appear. Conversations offer cues that geometry lacks, such as who is speaking and what is said, but no gaze dataset provides them for many multi-party scenes. We introduce SORBET, a benchmark of 3,340 samples from 1,578 clips with synchronized speech, transcripts, and attended person labels, covering more multi-party scenes than existing gaze datasets. Prior gaze models struggle to use these cues, which come in different forms, must be linked to the right person, and can conflict with gaze geometry. We propose Nunchi, a framework that constructs a text prompt from the transcript and the outputs of gaze and speaker models, links every cue to the candidates with shared labels, and lets a vision-language model (VLM) weigh gaze geometry against conversational cues to identify the attended person. Nunchi outperforms the strongest existing method by 4.8 percentage points. On SORBET, conversational cues help whereas additional visual cues do not, and the gains hold across three VLMs, with automatic transcripts, and on two existing datasets.

cs.CV↗

GRAZE: Text-Free Grasp Synthesis for Unseen Objects from Human Gaze

Text-conditioned hand-object interaction (HOI) generation provides intuitive semantic control, but language is an indirect way to specify where an interaction should occur on an object. Naming the part to grasp requires object-specific words, and naming the approach side requires a reference frame that a prompt lacks. We propose GRAZE (GRAsp from gaZE), which generates hand-object grasp motion from recorded gaze instead of object and part names. Its Geometry-Aware Gaze Encoding (GeoGaze) maps gaze rays to cues on object geometry, and cross-attention conditions a diffusion Transformer on these cues. We also introduce HOT3D-Grasp, a grasp-centered benchmark with object-disjoint folds built on HOT3D recordings. On unseen objects, GRAZE grasps the target part more often than the strongest text-conditioned baseline (49.3% vs. 43.6%), achieves the highest grasp success (66.3% vs. 46.6%), and transfers without fine-tuning to GazeHOI (45.3% vs. 33.8%). Gaze conveys the approach side: when gaze is rotated around the object, the approach turns with it in 59.1% of cases, against 20.5% for a text model retrained with clock words. On the same backbone, a text prompt names parts better, but gaze adds the approach side. Given both, GRAZE keeps the prompt's part accuracy and follows gaze.

cs.CV↗

Context-Aware Autoregressive Diffusion for Gloss-Wise Sign Language Production

To generate natural and accurate sentence-level sign language, synthesizing the "gloss", the fundamental semantic unit, is essential. However, most current sign-language production (SLP) methods generate entire sequences at once. While this end-to-end approach is often efficient, it is prone to temporal drift and hand motion blur as sentences get longer, and fails to accurately control individual glosses. In this paper, we propose the Context-aware Gloss-wise AutoRegressive Diffusion model (GARD), a gloss-wise diffusion framework that models coarticulation by conditioning on both semantic (linguistic) and kinematic (motion) contexts. To ensure natural continuity between gloss motions, GARD introduces two additional strategies: i) Inter-Gloss Transition Guidance, which applies gradient-based guidance to kinematically align inter-gloss boundaries and ensure seamless pose consistency. ii) Global Motion Harmonizer, refining the entire gloss motion sequence based on the boundary poses adjusted by Inter-Gloss Transition Guidance. Extensive experiments on Phoenix-T and CSL-Daily datasets demonstrate that GARD achieves superior performance over existing SLP methods in terms of both linguistic accuracy and motion similarity.

cs.CV↗

Roll Your Eyes: Gaze Redirection via Explicit 3D Eyeball Rotation

We propose a novel 3D gaze redirection framework that leverages an explicit 3D eyeball structure. Existing gaze redirection methods are typically based on neural radiance fields, which employ implicit neural representations via volume rendering. Unlike these NeRF-based approaches, where the rotation and translation of 3D representations are not explicitly modeled, we introduce a dedicated 3D eyeball structure to represent the eyeballs with 3D Gaussian Splatting (3DGS). Our method generates photorealistic images that faithfully reproduce the desired gaze direction by explicitly rotating and translating the 3D eyeball structure. In addition, we propose an adaptive deformation module that enables the replication of subtle muscle movements around the eyes. Through experiments conducted on the ETH-XGaze dataset, we demonstrate that our framework is capable of generating diverse novel gaze images, achieving superior image quality and gaze estimation accuracy compared to previous state-of-the-art methods.

cs.CV↗

PersonaBooth: Personalized Text-to-Motion Generation

This paper introduces Motion Personalization, a new task that generates personalized motions aligned with text descriptions using several basic motions containing Persona. To support this novel task, we introduce a new large-scale motion dataset called PerMo (PersonaMotion), which captures the unique personas of multiple actors. We also propose a multi-modal finetuning method of a pretrained motion diffusion model called PersonaBooth. PersonaBooth addresses two main challenges: i) A significant distribution gap between the persona-focused PerMo dataset and the pretraining datasets, which lack persona-specific data, and ii) the difficulty of capturing a consistent persona from the motions vary in content (action type). To tackle the dataset distribution gap, we introduce a persona token to accept new persona features and perform multi-modal adaptation for both text and visuals during finetuning. To capture a consistent persona, we incorporate a contrastive learning technique to enhance intra-cohesion among samples with the same persona. Furthermore, we introduce a context-aware fusion mechanism to maximize the integration of persona cues from multiple input motions. PersonaBooth outperforms state-of-the-art motion style transfer methods, establishing a new benchmark for motion personalization.

cs.CV↗

DIR-ST$^2$: Delineation of Imprecise Regions Using Spatio--Temporal--Textual Information

An imprecise region is referred to as a geographical area without a clearly-defined boundary in the literature. Previous clustering-based approaches exploit spatial information to find such regions. However, the prior studies suffer from the following two problems: the subjectivity in selecting clustering parameters and the inclusion of a large portion of the undesirable region (i.e., a large number of noise points). To overcome these problems, we present DIR-ST$^2$, a novel framework for delineating an imprecise region by iteratively performing density-based clustering, namely DBSCAN, along with not only spatio--textual information but also temporal information on social media. Specifically, we aim at finding a proper radius of a circle used in the iterative DBSCAN process by gradually reducing the radius for each iteration in which the temporal information acquired from all resulting clusters are leveraged. Then, we propose an efficient and automated algorithm delineating the imprecise region via hierarchical clustering. Experiment results show that by virtue of the significant noise reduction in the region, our DIR-ST$^2$ method outperforms the state-of-the-art approach employing one-class support vector machine in terms of the $\mathcal{F}_1$ score from comparison with precisely-defined regions regarded as a ground truth, and returns apparently better delineation of imprecise regions. The computational complexity of DIR-ST$^2$ is also analytically and numerically shown.

cs.IR↗