Synscapes: A Photorealistic Synthetic Dataset for Street Scene Parsing

TL;DR

Synscapes is a highly realistic synthetic dataset for street scene parsing that enables detailed analysis of computer vision models and improves training outcomes for real-world applications.

Contribution

The paper introduces Synscapes, a photorealistic synthetic dataset, and demonstrates its effectiveness for training and analyzing street scene understanding models.

Findings

State-of-the-art results in street scene parsing

Insights into model behavior on synthetic versus real data

Detailed analysis of factors affecting model performance

Abstract

We introduce Synscapes -- a synthetic dataset for street scene parsing created using photorealistic rendering techniques, and show state-of-the-art results for training and validation as well as new types of analysis. We study the behavior of networks trained on real data when performing inference on synthetic data: a key factor in determining the equivalence of simulation environments. We also compare the behavior of networks trained on synthetic data and evaluated on real-world data. Additionally, by analyzing pre-trained, existing segmentation and detection models, we illustrate how uncorrelated images along with a detailed set of annotations open up new avenues for analysis of computer vision systems, providing fine-grain information about how a model's performance changes according to factors such as distance, occlusion and relative object orientation.