Jiajun Wu

We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. Expand

We propose to learn task-oriented flow (TOFlow) by performing motion analysis and video processing jointly in an end-to-end trainable convolutional network. Expand

We present Pix3D, a large-scale benchmark of diverse image-shape pairs with pixel-level 2D-3D alignment. Expand

We propose 3D INterpreter Network (3D-INN), an end-to-end framework for recovering 3D object skeletons, trained on both real 2D-labeled images and synthetic 3D objects. Expand

We study the problem of synthesizing a number of likely future frames from a single input image. Expand

In this paper, we attempt to model deep learning in a weakly supervised learning framework and apply the learned visual knowledge to assist the task of image classification. Expand

We formulate the interactive segmentation problem as a multiple instance learning (MIL) task by generating positive bags from pixels of sweeping lines within a bounding box. Expand

We marry two powerful ideas: deep representation learning for visual recognition and language understanding, and symbolic program execution for reasoning, proposing a neural-symbolic approach for visual question answering. Expand

We propose MarrNet, an end-to-end trainable model that sequentially estimates 2.5D sketches and 3D object shape. Expand

We propose a generative model for solving these problems of physical scene understanding from real-world videos and images. Expand