Hyeonwoo Kim

Inwoo Hwang, Hyeonwoo Kim, Young Min Kim

We propose Text2Scene, a method to automatically create realistic textures for virtual scenes composed of multiple objects. Guided by a reference image and text descriptions, our pipeline adds detailed texture on labeled 3D geometries in the room such that the generated colors respect the hierarchical structure or semantic parts that are often composed of similar materials. Instead of applying flat stylization on the entire scene at a single step, we obtain weak semantic cues from geometric segmentation, which are further clarified by assigning initial colors to segmented parts. Then we add texture details for individual objects such that their projections on image space exhibit feature embedding aligned with the embedding of the input. The decomposition makes the entire pipeline tractable to a moderate amount of computation resources and memory. As our framework utilizes the existing resources of image and text embedding, it does not require dedicated datasets with high-quality textures designed by skillful artists. To the best of our knowledge, it is the first practical and scalable approach that can create detailed and realistic textures of the desired style that maintain structural context for scenes with multiple objects.

Hyeonwoo Kim, Sangwon Baik, Hanbyul Joo

Modeling how humans interact with objects is crucial for AI to effectively assist or mimic human behaviors. Existing studies for learning such ability primarily focus on static human-object interaction (HOI) patterns, such as contact and spatial relationships, while dynamic HOI patterns, capturing the movement of humans and objects over time, remain relatively underexplored. In this paper, we present a novel framework for learning Dynamic Affordance across various target object categories. To address the scarcity of 4D HOI datasets, our method learns the 3D dynamic affordance from synthetically generated 4D HOI samples. Specifically, we propose a pipeline that first generates 2D HOI videos from a given 3D target object using a pre-trained video diffusion model, then lifts them into 3D to generate 4D HOI samples. Leveraging these synthesized 4D HOI samples, we train DAViD, our generative 4D human-object interaction model, which is composed of two key components: (1) a human motion diffusion model (MDM) with Low-Rank Adaptation (LoRA) module to fine-tune a pre-trained MDM to learn the HOI motion concepts from limited HOI motion samples, (2) a motion diffusion model for 4D object poses conditioned by produced human interaction motions. Interestingly, DAViD can integrate newly learned HOI motion concepts with pre-trained human motions to create novel HOI motions, even for multiple HOI motion concepts, demonstrating the advantage of our pipeline with LoRA in integrating dynamic HOI concepts. Through extensive experiments, we demonstrate that DAViD outperforms baselines in synthesizing HOI motion.

Dahyun Kim, Chanjun Park, Sanghoon Kim et al.

We introduce SOLAR 10.7B, a large language model (LLM) with 10.7 billion parameters, demonstrating superior performance in various natural language processing (NLP) tasks. Inspired by recent efforts to efficiently up-scale LLMs, we present a method for scaling LLMs called depth up-scaling (DUS), which encompasses depthwise scaling and continued pretraining. In contrast to other LLM up-scaling methods that use mixture-of-experts, DUS does not require complex changes to train and inference efficiently. We show experimentally that DUS is simple yet effective in scaling up high-performance LLMs from small ones. Building on the DUS model, we additionally present SOLAR 10.7B-Instruct, a variant fine-tuned for instruction-following capabilities, surpassing Mixtral-8x7B-Instruct. SOLAR 10.7B is publicly available under the Apache 2.0 license, promoting broad access and application in the LLM field.

Hyungjoon Kim, Hyeonwoo Kim, Eenjun Hwang

Detection of facial landmarks and accurate tracking of their shape are essential in real-time virtual makeup applications, where users can see the makeups effect by moving their face in different directions. Typical face tracking techniques detect diverse facial landmarks and track them using a point tracker such as the Kanade-Lucas-Tomasi (KLT) point tracker. Typically, 5 or 64 points are used for tracking a face. Even though these points are sufficient to track the approximate locations of facial landmarks, they are not sufficient to track the exact shape of facial landmarks. In this paper, we propose a method that can track the exact shape of facial landmarks in real-time by combining a deep learning technique and a point tracker. We detect facial landmarks accurately using SegNet, which performs semantic segmentation based on deep learning. Edge points of detected landmarks are tracked using the KLT point tracker. In spite of its popularity, the KLT point tracker suffers from the point loss problem. We solve this problem by executing SegNet periodically to calculate the shape of facial landmarks. That is, by combining the two techniques, we can avoid the computational overhead of SegNet for real-time shape tracking and the point loss problem of the KLT point tracker. We performed several experiments to evaluate the performance of our method and report some of the results herein.