Search Results for author:
Found 7 papers, 3 papers with code
MotionChain: Conversational Motion Controllers via Multimodal Prompts
However, this proficiency remains largely unexplored in other multimodal generative models, particularly in human motion models.
Object Segmentation-Assisted Inter Prediction for Versatile Video Coding
Moreover, segmentation mask is considered in the joint rate-distortion optimization for motion estimation and partition estimation to derive the motion vector of different regions and partition more accurately.
Latent assimilation with implicit neural representations for unknown dynamics
Data assimilation is crucial in a wide range of applications, but it often faces challenges such as high computational costs due to data dimensionality and incomplete understanding of underlying mechanisms.
Learning to simulate partially known spatio-temporal dynamics with trainable difference operators
Recently, using neural networks to simulate spatio-temporal dynamics has received a lot of attention.
Offline and Online Optical Flow Enhancement for Deep Video Compression
In the offline stage, we fine-tune a trained optical flow estimation network with the motion information provided by a traditional (non-deep) video compression scheme, e. g. H. 266/VVC, as we believe the motion information of H. 266/VVC achieves a better rate-distortion trade-off.
Transformers in Medical Image Analysis: A Review
Transformers have dominated the field of natural language processing, and recently impacted the computer vision area.
Cross-Modality Brain Tumor Segmentation via Bidirectional Global-to-Local Unsupervised Domain Adaptation
Specifically, a bidirectional image synthesis and segmentation module is proposed to segment the brain tumor using the intermediate data distributions generated for the two domains, which includes an image-to-image translator and a shared-weighted segmentation network.
