DepthSynth: Real-Time Realistic Synthetic Data Generation from CAD Models for 2.5D Recognition
Authors:
Benjamin Planche,
Ziyan Wu,
Kai Ma,
Shanhui Sun,
Stefan Kluckner,
Terrence Chen,
Andreas Hutter,
Sergey Zakharov,
Harald Kosch,
Jan Ernst
Abstract:
Recent progress in computer vision has been dominated by deep neural networks trained over large amounts of labeled data. Collecting such datasets is however a tedious, often impossible task; hence a surge in approaches relying solely on synthetic data for their training. For depth images however, discrepancies with real scans still noticeably affect the end performance. We thus propose an end-to-…
▽ More
Recent progress in computer vision has been dominated by deep neural networks trained over large amounts of labeled data. Collecting such datasets is however a tedious, often impossible task; hence a surge in approaches relying solely on synthetic data for their training. For depth images however, discrepancies with real scans still noticeably affect the end performance. We thus propose an end-to-end framework which simulates the whole mechanism of these devices, generating realistic depth data from 3D models by comprehensively modeling vital factors e.g. sensor noise, material reflectance, surface geometry. Not only does our solution cover a wider range of sensors and achieve more realistic results than previous methods, assessed through extended evaluation, but we go further by measuring the impact on the training of neural networks for various recognition tasks; demonstrating how our pipeline seamlessly integrates such architectures and consistently enhances their performance.
△ Less
Submitted 28 November, 2017; v1 submitted 27 February, 2017;
originally announced February 2017.
Geometric Abstraction from Noisy Image-Based 3D Reconstructions
Authors:
Thomas Holzmann,
Christof Hoppe,
Stefan Kluckner,
Horst Bischof
Abstract:
Creating geometric abstracted models from image-based scene reconstructions is difficult due to noise and irregularities in the reconstructed model. In this paper, we present a geometric modeling method for noisy reconstructions dominated by planar horizontal and orthogonal vertical structures. We partition the scene into horizontal slices and create an inside/outside labeling represented by a flo…
▽ More
Creating geometric abstracted models from image-based scene reconstructions is difficult due to noise and irregularities in the reconstructed model. In this paper, we present a geometric modeling method for noisy reconstructions dominated by planar horizontal and orthogonal vertical structures. We partition the scene into horizontal slices and create an inside/outside labeling represented by a floor plan for each slice by solving an energy minimization problem. Consecutively, we create an irregular discretization of the volume according to the individual floor plans and again label each cell as inside/outside by minimizing an energy function. By adjusting the smoothness parameter, we introduce different levels of detail. In our experiments, we show results with varying regularization levels using synthetically generated and real-world data.
△ Less
Submitted 19 April, 2014;
originally announced April 2014.
