Diverse Motion In-betweening from Sparse Keyframes with Dual Posture Stitching.

In-betweening is a technique for generating transitions given start and target character states. The majority of existing works require multiple (often ≥ 10) frames as input, which are not always available. In addition, they produce results that lack diversity, which may not fulfill artists' requirements. Addressing these gaps, our work deals with a focused yet challenging problem: generating diverse and high-quality transitions given exactly two frames (only the start and target frames). To cope with this challenging scenario, we propose a bi-directional motion generation and stitching scheme which generates forward and backward transitions from the start and target frames with two adversarial autoregressive networks, respectively, and stitches them midway between the start and target frames. In contrast to stitching at the start or target frames, where the ground truth cannot be altered, there is no strict midway ground truth. Thus, our method can capitalize on this flexibility and generate high-quality and diverse transitions simultaneously. Specifically, we employ conditional variational autoencoders (CVAEs) to implement our autoregressive networks and propose a novel stitching loss to stitch the bi-directional generated motions around the midway point.

Droplet-Transmitted Infection Risk Ranking Based on Close Proximity Interaction.

We propose an automatic method to identify people who are potentially-infected by droplet-transmitted diseases. This high-risk group of infection was previously identified by conducting large-scale visits/interviews, or manually screening among tons of recorded surveillance videos. Both are time-intensive and most likely to delay the control of communicable diseases like influenza. In this paper, we address this challenge by solving a multi-tasking problem from the captured surveillance videos. This multi-tasking framework aims to model the principle of Close Proximity Interaction and thus infer the infection risk of individuals. The complete workflow includes three essential sub-tasks: (1) person re-identification (REID), to identify the diagnosed patient and infected individuals across different cameras, (2) depth estimation, to provide a spatial knowledge of the captured environment, (3) pose estimation, to evaluate the distance between the diagnosed and potentially-infected subjects. Our method significantly reduces the time and labor costs. We demonstrate the advantages of high accuracy and efficiency of our method. Our method is expected to be effective in accelerating the process of identifying the potentially infected group and ultimately contribute to the well-being of public health.