An Evaluation of Synthetic Data for Deep Learning Stereo Depth Algorithms
- Resource Type
- Authors
- David Moloney; Kevin J. Lee
- Source
- Proceedings of the International Conference on Watermarking and Image Processing.
- Subject
- Ground truth
Data acquisition
Stereopsis
Margin (machine learning)
business.industry
Computer science
Deep learning
Artificial intelligence
business
Convolutional neural network
Algorithm
Field (computer science)
Synthetic data
- Language
Stereo vision is a very active field in the realm of computer vision and in recent years Convolutional Neural Networks (CNNs) have proven to be very competitive against the state-of-the-art. However, the performance of these networks are limited by the quality of the data that is used when training the CNNs. Data acquisition of high quality labeled images is a time-consuming and expensive process. By exploiting the power of modern-day powerful GPUs, we present a synthetic data-set with fully rectified stereo image pairs and accompanying accurate ground truth information that can be used for training and testing stereo algorithms. We provide validation of the quality of our dataset by performing quantitative experiments that suggest pre-training deep learning algorithms on synthetic data can perform competitively against networks trained on real life data. Testing on a real-world data-set, we found the accuracy performance difference between the real and synthetically trained networks was within a margin of 1.8%. We also illustrate the functionality synthetic data can provide, by conducting a key performance index on a selection of conventional and deep learning stereo algorithms available on embedded platforms and compared them under common metrics.