In this paper, we introduce a novel framework for creating multimodal interactive digital twin characters, from dialogue videos of TV shows. Specifically, these digital twin characters are capable of responding to user inputs with harmonious textual, vocal, and visual content. They not only replicate the external characteristics, such as appearance and tone, but also capture internal attributes, including personality and habitual behaviors. To support this ambitious task, we collect the Multimodal Character-Centric Conversation Dataset, named MCCCD, which includes character-specific and high-quality multimodal dialogue data with detailed annotations, featuring 6.8 k utterances and 4.6 hours of audio/video per character. Notably, the MCCCD dataset is approximately ten times larger than existing datasets in terms of per-character data volume, facilitating the detailed modeling of complex character-centric traits. Further, we propose a baseline framework to create digital twin characters, consists of dialogue generation through large language models, voice generation via speech synthesis models, and visual representation with 3D talking head models. Experimental results demonstrate that our approach significantly outperforms existing methods in generating consistent and character-specific responses, setting a new benchmark for digital character creation. Our collected dataset and proposed baseline have paved the way for the creation of highly interactive and natural digital avatars, opening the door to extensive and practical applications of digital humans.

Segmenting medical images into different tissues is an important task in medical image analysis, e.g., classifying every voxel of input image into different tissue types: CSF, gray matter and white matter. This paper investigates the fully-tuned radial basis function (RBF) and compares it with the traditional fuzzy c-mean (FCM) clustering algorithm in MR image segmentation. It turns out that FCM is not only biased by the number of voxels in different groups, but also by the intensity differences between different tissue groups, while the fully-tuned RBF captures the multi-Gaussian distribution of the image intensities very well and thus it can be used to segment image intensities accurately. Moreover, in order to generate spatially smooth segmentation results, a Markov random field model is applied to the segmentation results of the fully-tuned RBF algorithm. Experimental results show that fully-tuned RBF method can capture the tissue intensity distribution more accurately than the FCM algorithm