Recent advancements in deep learning have shown significant potential in developing barrier-free environments for the hearing-impaired. However, the lack of diverse and specialized datasets limits the development of high-level services such as interactive museum exhibitions. In this paper, we introduce KSL-Ex, a Korean Sign Language (KSL) dataset designed to support real-world interactive exhibition services. Our dataset comprises 29,574 sign language video samples, including isolated words, continuous KSL sentences, questionanswer pairs, and detailed annotations. In addition, we propose an interactive sign language question answering framework that leverages state-of-the-art large language models which have demonstrated outstanding performance in question answering tasks. The proposed framework consists of four key components: The sign language recognition (SLR) model, the answering module, the refinement module, and the 3 D sign language animation module. These components effectively process sign language queries, generate proper answers, and produce sign language animations to provide interactive communication interfaces for the hearing-impaired. Experimental results show the effectiveness of both KSL-Ex and the proposed framework, demonstrating their potential in providing real-world interactive sign language services.

Recent significant advancements in diffusion models have revolutionized image generation, enabling the synthesis of highly realistic images with text-based guidance. These breakthroughs have paved the way for constructing datasets via generative artificial intelligence (AI), offering immense potential for various applications. However, two critical challenges hinder the widespread adoption of synthesized data: computational cost and the generation of peculiar images. While computational costs have improved through various approaches, the issue of peculiar image generation remains relatively unexplored. Existing solutions rely on heuristics, extra training, or AI-based post-processing to mitigate this problem. In this paper, we present a novel approach to address both issues simultaneously. We establish that both gradient descent and diffusion sampling are specific cases of the generalized expectation maximization algorithm. We hypothesize and empirically demonstrate that peculiar image generation is akin to the local minima problem in optimization. Inspired by optimization techniques, we apply naive momentum and positive-negative momentum to diffusion sampling. Last, we propose new metrics to evaluate the peculiarity. Experimental results show momentum effectively prevents peculiar image generation without extra computation.