Junhao Zhang - Academia.edu (original) (raw)

Yogesh Rawat related author profile picture

Sarah Shiraz related author profile picture

Shruti Vyas related author profile picture

Hochul Hwang related author profile picture

Srishti Goel related author profile picture

Sandika Biswas related author profile picture

Julien Valentin related author profile picture

Ankan Dash related author profile picture

Wesley Beccaro related author profile picture

Miguel  Arjona Ramírez related author profile picture

Uploads

Papers by Junhao Zhang

Research paper thumbnail of Pose-guided Generative Adversarial Net for Novel View Action Synthesis

ArXiv, 2021

We focus on the problem of novel-view human action synthesis. Given an action video, the goal is ... more We focus on the problem of novel-view human action synthesis. Given an action video, the goal is to generate the same action from an unseen viewpoint. Naturally, novel view video synthesis is more challenging than image synthesis. It requires the synthesis of a sequence of realistic frames with temporal coherency. Besides, transferring different actions to a novel target view requires awareness of action category and viewpoint change simultaneously. To address these challenges we propose a novel framework named Pose-guided Action Separable Generative Adversarial Net (PAS-GAN), which utilizes pose to alleviate the difficulty of this task. First, we propose a recurrent posetransformation module which transforms actions from the source view to the target view and generates novel view pose sequence in 2D coordinate space. Second, a welltransformed pose sequence enables us to separate the action and background in the target view. We employ a novel local-global spatial transformation modu...

Research paper thumbnail of Pose-guided Generative Adversarial Net for Novel View Action Synthesis

ArXiv, 2021

We focus on the problem of novel-view human action synthesis. Given an action video, the goal is ... more We focus on the problem of novel-view human action synthesis. Given an action video, the goal is to generate the same action from an unseen viewpoint. Naturally, novel view video synthesis is more challenging than image synthesis. It requires the synthesis of a sequence of realistic frames with temporal coherency. Besides, transferring different actions to a novel target view requires awareness of action category and viewpoint change simultaneously. To address these challenges we propose a novel framework named Pose-guided Action Separable Generative Adversarial Net (PAS-GAN), which utilizes pose to alleviate the difficulty of this task. First, we propose a recurrent posetransformation module which transforms actions from the source view to the target view and generates novel view pose sequence in 2D coordinate space. Second, a welltransformed pose sequence enables us to separate the action and background in the target view. We employ a novel local-global spatial transformation modu...

Log In