OmniFlow: Human Omnidirectional Optical Flow

TL;DR

OmniFlow introduces a comprehensive synthetic dataset of omnidirectional human optical flow in indoor environments, facilitating improved training of neural networks for optical flow estimation.

Contribution

The paper presents OmniFlow, a novel large-scale synthetic dataset with realistic indoor scenes and human motions for optical flow training.

Findings

Training on OmniFlow improves optical flow estimation accuracy.

Test-Time-Augmentation enhances model performance.

Dataset includes 23,653 image pairs with optical flow annotations.

Abstract

Optical flow is the motion of a pixel between at least two consecutive video frames and can be estimated through an end-to-end trainable convolutional neural network. To this end, large training datasets are required to improve the accuracy of optical flow estimation. Our paper presents OmniFlow: a new synthetic omnidirectional human optical flow dataset. Based on a rendering engine we create a naturalistic 3D indoor environment with textured rooms, characters, actions, objects, illumination and motion blur where all components of the environment are shuffled during the data capturing process. The simulation has as output rendered images of household activities and the corresponding forward and backward optical flow. To verify the data for training volumetric correspondence networks for optical flow estimation we train different subsets of the data and test on OmniFlow with and without Test-Time-Augmentation. As a result we have generated 23,653 image pairs and corresponding forward and backward optical flow. Our dataset can be downloaded from: https://mytuc.org/byfs