Tae-Hwan Kim

Jinsik Bang, Jaeyeon Bae, Donggyu Lee et al.

Vision-language models (VLMs) have shown strong perception and reasoning abilities for instruction-following embodied agents. However, despite these abilities and their generalization performance, they still face limitations in environmental understanding, often failing on interactions or relying on environment metadata during execution. To address this challenge, we propose a novel framework named Environmental Understanding Embodied Agent (EUEA), which fine-tunes four core skills: 1) object perception for identifying relevant objects, 2) task planning for generating interaction subgoals, 3) action understanding for judging success likelihood, and 4) goal recognition for determining goal completion. By fine-tuning VLMs with EUEA skills, our framework enables more reliable task execution for instruction-following. We further introduce a recovery step that leverages these core skills and a group relative policy optimization (GRPO) stage that refines inconsistent skill predictions. The recovery step samples alternative actions to correct failure cases, and the GRPO stage refines inconsistent skill predictions. Across ALFRED tasks, our VLM significantly outperforms a behavior-cloning baseline, achieving an 8.86% improvement in average success rate. The recovery and GRPO stages provide an additional 3.03% gain, further enhancing overall performance. Finally, our skill-level analyses reveal key limitations in the environmental understanding of closed- and open-source VLMs and identify the capabilities necessary for effective agent-environment interaction.

Cheng-Wei Ching, Xin Chen, Taehwan Kim et al.

Federated Learning (FL) is an emerging distributed machine learning (ML) technique that enables in-situ model training and inference on decentralized edge devices. We propose Totoro$^+$, a novel scalable FL system that enables massive FL applications to run simultaneously on edge networks. The key insight is to explore a distributed hash table (DHT)-based peer-to-peer (P2P) model to re-architect the centralized FL system design into a fully decentralized one. In contrast to previous studies where many FL applications shared one centralized parameter server, Totoro$^+$ assigns a dedicated parameter server to each application. Any edge node can act as any application's coordinator, aggregator, client selector, worker (participant device), or any combination of the above, thereby radically improving scalability and adaptivity. Totoro$^+$ introduces three innovations to realize its design: a locality-aware P2P multi-ring structure, a publish/subscribe-based forest abstraction, and a game-theoretic path planning model with a guarantee of an $ε$-approximate Nash equilibrium. Real-world experiments on 500 Amazon EC2 servers show that Totoro$^+$ scales gracefully with the number of FL applications and $N$ edge nodes speeds up the total training time by $1.2\times-14.0\times$, achieves $\mathcal{O}(\log N)$ hops for model dissemination and gradient aggregation with millions of nodes, and efficiently adapts to the practical edge networks and churns.

Chanhyuk Choi, Taesoo Kim, Donggyu Lee et al.

Talking face generation has gained significant attention as a core application of generative models. To enhance the expressiveness and realism of synthesized videos, emotion editing in talking face video plays a crucial role. However, existing approaches often limit expressive flexibility and struggle to generate extended emotions. Label-based methods represent emotions with discrete categories, which fail to capture a wide range of emotions. Audio-based methods can leverage emotionally rich speech signals - and even benefit from expressive text-to-speech (TTS) synthesis - but they fail to express the target emotions because emotions and linguistic contents are entangled in emotional speeches. Images-based methods, on the other hand, rely on target reference images to guide emotion transfer, yet they require high-quality frontal views and face challenges in acquiring reference data for extended emotions (e.g., sarcasm). To address these limitations, we propose Cross-Modal Emotion Transfer (C-MET), a novel approach that generates facial expressions based on speeches by modeling emotion semantic vectors between speech and visual feature spaces. C-MET leverages a large-scale pretrained audio encoder and a disentangled facial expression encoder to learn emotion semantic vectors that represent the difference between two different emotional embeddings across modalities. Extensive experiments on the MEAD and CREMA-D datasets demonstrate that our method improves emotion accuracy by 14% over state-of-the-art methods, while generating expressive talking face videos - even for unseen extended emotions. Code, checkpoint, and demo are available at https://chanhyeok-choi.github.io/C-MET/

Hyeonyu Kim, Jongeun Kim, Jeonghun Kang et al.

This technical report presents the 2nd winning model for AQTC, a task newly introduced in CVPR 2022 LOng-form VidEo Understanding (LOVEU) challenges. This challenge faces difficulties with multi-step answers, multi-modal, and diverse and changing button representations in video. We address this problem by proposing a new context ground module attention mechanism for more effective feature mapping. In addition, we also perform the analysis over the number of buttons and ablation study of different step networks and video features. As a result, we achieved the overall 2nd place in LOVEU competition track 3, specifically the 1st place in two out of four evaluation metrics. Our code is available at https://github.com/jaykim9870/ CVPR-22_LOVEU_unipyler.

Jongeun Kim, MinChung Kim, Taehwan Kim

Slogans play a crucial role in building the brand's identity of the firm. A slogan is expected to reflect firm's vision and brand's value propositions in memorable and likeable ways. Automating the generation of slogans with such characteristics is challenging. Previous studies developted and tested slogan generation with syntactic control and summarization models which are not capable of generating distinctive slogans. We introduce a a novel apporach that leverages pre-trained transformer T5 model with noise perturbation on newly proposed 1:N matching pair dataset. This approach serves as a contributing fator in generting distinctive and coherent slogans. Turthermore, the proposed approach incorporates descriptions about the firm and brand into the generation of slogans. We evaluate generated slogans based on ROUGE1, ROUGEL and Cosine Similarity metrics and also assess them with human subjects in terms of slogan's distinctiveness, coherence, and fluency. The results demonstrate that our approach yields better performance than baseline models and other transformer-based models.

Taegyeong Lee, Jeonghun Kang, Hyeonyu Kim et al.

Representing wild sounds as images is an important but challenging task due to the lack of paired datasets between sound and images and the significant differences in the characteristics of these two modalities. Previous studies have focused on generating images from sound in limited categories or music. In this paper, we propose a novel approach to generate images from in-the-wild sounds. First, we convert sound into text using audio captioning. Second, we propose audio attention and sentence attention to represent the rich characteristics of sound and visualize the sound. Lastly, we propose a direct sound optimization with CLIPscore and AudioCLIP and generate images with a diffusion-based model. In experiments, it shows that our model is able to generate high quality images from wild sounds and outperforms baselines in both quantitative and qualitative evaluations on wild audio datasets.

Taegyeong Lee, Jinsik Bang, Soyeong Kwon et al.

Recent advancements in deep learning have significantly improved performance on computer vision tasks. Previous image classification methods primarily modify model architectures or add features, and they optimize models using cross-entropy loss on class logits. Since they focus on classifying images with considering class labels, these methods may struggle to learn various \emph{aspects} of classes (e.g., natural positions and shape changes). Rethinking the previous approach from a novel view, we propose a multi-aspect knowledge distillation method using Multimodal Large Language Models (MLLMs). Our approach involves: 1) querying Large Language Model with multi-aspect questions relevant to the knowledge we want to transfer to the model, 2) extracting corresponding logits from MLLM, and 3) expanding the model's output dimensions to distill these multi-aspect logits. We then apply cross-entropy loss to class logits and binary cross-entropy loss to multi-aspect logits. Through our method, the model can learn not only the knowledge about visual aspects but also the abstract and complex aspects that require a deeper understanding. We primarily apply our method to image classification, and to explore the potential for extending our model, such as object detection. In all experimental results, our method improves the performance of the baselines. Additionally, we analyze the effect of multi-aspect knowledge distillation. These results demonstrate that our method can transfer knowledge about various aspects to the model and the aspect knowledge can enhance model performance in computer vision tasks.

Junha Song, Yongsik Jo, So Yeon Min et al. · cmu

Image captioning is fundamental for applications like video-grounded chatbot systems and navigation robots, yet deploying such models on local devices is challenging due to the high computational demands of multimodal LLMs (MLLMs). To address this, we first build lightweight captioning models using a 125M-parameter language model, 56 times smaller than LLaMA-7B, and evaluate their performance not only on single-sentence but on detailed captioning tasks. We obtain surprising results showing that our model can achieve performance comparable to MLLMs, suggesting its potential to serve as a strong captioning specialist for on-device applications. While promising, our model also exhibits a limitation: like other MLLMs, it suffers from occasional captioning errors. We investigate the underlying causes and observe that the problems stem from ineffective attention mechanisms and limited visual representations. To alleviate them, we develop a novel captioning framework, Sharp-Eyed Refinement, which enhances caption quality by refining coarse descriptions into more precise captions. At its core, DeepLens improves visual grounding by re-examining the informative regions identified in the initial glance. Experimental results demonstrate the superiority of our model over both recent lightweight captioning methods and MLLMs in detailed captioning and even in long-range video QA tasks.

Hyeonyu Kim, Seokhoon Jeong, Seonghee Han et al.

Visual editing with diffusion models has made significant progress but often struggles with complex scenarios that textual guidance alone could not adequately describe, highlighting the need for additional non-text editing prompts. In this work, we introduce a novel audio-guided visual editing framework that can handle complex editing tasks with multiple text and audio prompts without requiring additional training. Existing audio-guided visual editing methods often necessitate training on specific datasets to align audio with text, limiting their generalization to real-world situations. We leverage a pre-trained multi-modal encoder with strong zero-shot capabilities and integrate diverse audio into visual editing tasks, by alleviating the discrepancy between the audio encoder space and the diffusion model's prompt encoder space. Additionally, we propose a novel approach to handle complex scenarios with multiple and multi-modal editing prompts through our separate noise branching and adaptive patch selection. Our comprehensive experiments on diverse editing tasks demonstrate that our framework excels in handling complicated editing scenarios by incorporating rich information from audio, where text-only approaches fail.

Hansoo Park, Chanwoo Kim, Jihyeon Kim et al.

RGB-based 3D pose estimation methods have been successful with the development of deep learning and the emergence of high-quality 3D pose datasets. However, most existing methods do not operate well for testing images whose distribution is far from that of training data. However, most existing methods do not operate well for testing images whose distribution is far from that of training data. This problem might be alleviated by involving diverse data during training, however it is non-trivial to collect such diverse data with corresponding labels (i.e. 3D pose). In this paper, we introduced an unsupervised domain adaptation framework for 3D pose estimation that utilizes the unlabeled data in addition to labeled data via masked image modeling (MIM) framework. Foreground-centric reconstruction and attention regularization are further proposed to increase the effectiveness of unlabeled data usage. Experiments are conducted on the various datasets in human and hand pose estimation tasks, especially using the cross-domain scenario. We demonstrated the effectiveness of ours by achieving the state-of-the-art accuracy on all datasets.

Xudong Liu, Tao Li, Hao Peng et al.

The concept of beauty has been debated by philosophers and psychologists for centuries, but most definitions are subjective and metaphysical, and deficit in accuracy, generality, and scalability. In this paper, we present a novel study on mining beauty semantics of facial attributes based on big data, with an attempt to objectively construct descriptions of beauty in a quantitative manner. We first deploy a deep convolutional neural network (CNN) to extract facial attributes, and then investigate correlations between these features and attractiveness on two large-scale datasets labelled with beauty scores. Not only do we discover the secrets of beauty verified by statistical significance tests, our findings also align perfectly with existing psychological studies that, e.g., small nose, high cheekbones, and femininity contribute to attractiveness. We further leverage these high-level representations to original images by a generative adversarial network (GAN). Beauty enhancements after synthesis are visually compelling and statistically convincing verified by a user survey of 10,000 data points.

Seyed Hamidreza Mohammadi, Taehwan Kim

We study the problem of cross-lingual voice conversion in non-parallel speech corpora and one-shot learning setting. Most prior work require either parallel speech corpora or enough amount of training data from a target speaker. However, we convert an arbitrary sentences of an arbitrary source speaker to target speaker's given only one target speaker training utterance. To achieve this, we formulate the problem as learning disentangled speaker-specific and context-specific representations and follow the idea of [1] which uses Factorized Hierarchical Variational Autoencoder (FHVAE). After training FHVAE on multi-speaker training data, given arbitrary source and target speakers' utterance, we estimate those latent representations and then reconstruct the desired utterance of converted voice to that of target speaker. We investigate the effectiveness of the approach by conducting voice conversion experiments with varying size of training utterances and it was able to achieve reasonable performance with even just one training utterance. We also examine the speech representation and show that World vocoder outperforms Short-time Fourier Transform (STFT) used in [1]. Finally, in the subjective tests, for one language and cross-lingual voice conversion, our approach achieved significantly better or comparable results compared to VAE-STFT and GMM baselines in speech quality and similarity.

Long Sha, Patrick Lucey, Stephan Zheng et al.

We propose a novel method for effective retrieval of multi-agent spatiotemporal tracking data. Retrieval of spatiotemporal tracking data offers several unique challenges compared to conventional text-based retrieval settings. Most notably, the data is fine-grained meaning that the specific location of agents is important in describing behavior. Additionally, the data often contains tracks of multiple agents (e.g., multiple players in a sports game), which generally leads to a permutational alignment problem when performing relevance estimation. Due to the frequent position swap of agents, it is difficult to maintain the correspondence of agents, and such issues make the pairwise comparison problematic for multi-agent spatiotemporal data. To address this issue, we propose a tree-based method to estimate the relevance between multi-agent spatiotemporal tracks. It uses a hierarchical structure to perform multi-agent data alignment and partitioning in a coarse-to-fine fashion. We validate our approach via user studies with domain experts. Our results show that our method boosts performance in retrieving similar sports plays -- especially in interactive situations where the user selects a subset of trajectories compared to current state-of-the-art methods.

Taehwan Kim, Jonathan Keane, Weiran Wang et al.

We study the problem of recognizing video sequences of fingerspelled letters in American Sign Language (ASL). Fingerspelling comprises a significant but relatively understudied part of ASL. Recognizing fingerspelling is challenging for a number of reasons: It involves quick, small motions that are often highly coarticulated; it exhibits significant variation between signers; and there has been a dearth of continuous fingerspelling data collected. In this work we collect and annotate a new data set of continuous fingerspelling videos, compare several types of recognizers, and explore the problem of signer variation. Our best-performing models are segmental (semi-Markov) conditional random fields using deep neural network-based features. In the signer-dependent setting, our recognizers achieve up to about 92% letter accuracy. The multi-signer setting is much more challenging, but with neural network adaptation we achieve up to 83% letter accuracies in this setting.

Taehwan Kim

In this thesis, we study the problem of recognizing video sequences of fingerspelled letters in American Sign Language (ASL). Fingerspelling comprises a significant but relatively understudied part of ASL, and recognizing it is challenging for a number of reasons: It involves quick, small motions that are often highly coarticulated; it exhibits significant variation between signers; and there has been a dearth of continuous fingerspelling data collected. In this work, we propose several types of recognition approaches, and explore the signer variation problem. Our best-performing models are segmental (semi-Markov) conditional random fields using deep neural network-based features. In the signer-dependent setting, our recognizers achieve up to about 8% letter error rates. The signer-independent setting is much more challenging, but with neural network adaptation we achieve up to 17% letter error rates.

Taehwan Kim, Weiran Wang, Hao Tang et al.

We study the problem of recognition of fingerspelled letter sequences in American Sign Language in a signer-independent setting. Fingerspelled sequences are both challenging and important to recognize, as they are used for many content words such as proper nouns and technical terms. Previous work has shown that it is possible to achieve almost 90% accuracies on fingerspelling recognition in a signer-dependent setting. However, the more realistic signer-independent setting presents challenges due to significant variations among signers, coupled with the dearth of available training data. We investigate this problem with approaches inspired by automatic speech recognition. We start with the best-performing approaches from prior work, based on tandem models and segmental conditional random fields (SCRFs), with features based on deep neural network (DNN) classifiers of letters and phonological features. Using DNN adaptation, we find that it is possible to bridge a large part of the gap between signer-dependent and signer-independent performance. Using only about 115 transcribed words for adaptation from the target signer, we obtain letter accuracies of up to 82.7% with frame-level adaptation labels and 69.7% with only word labels.