In this work, we investigate regression into the latent space as a probe to understand the compositional properties of GANs.
We show that, under a simple nonlinear operation, the data distribution can be modeled as Gaussian and therefore expressed using sufficient statistics.
The difficulty of annotating training data is a major obstacle to using CNNs for low-level tasks in video.
We address the unsupervised learning of several interconnected problems in low-level vision: single view depth prediction, camera motion estimation, optical flow, and segmentation of a video into the static scene and moving regions.
Ranked #30 on Monocular Depth Estimation on KITTI Eigen split
Existing optical flow datasets are limited in size and variability due to the difficulty of capturing dense ground truth.
Existing algorithms typically focus on either recovering motion and structure under the assumption of a purely static world or optical flow for general unconstrained scenes.
Ranked #5 on Optical Flow Estimation on Sintel-clean
Layered models allow scene segmentation and motion estimation to be formulated together and to inform one another.