Synscapes: A Photorealistic Synthetic Dataset for Street Scene Parsing

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.

arXiv commentFor more information and download see: https://7dlabs.com

References in corpus (2)

• Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation
• Procedural Modeling and Physically Based Rendering for Synthetic Data Generation in Automotive Applications

Cited by in corpus (32)

• Deep Multi-modal Object Detection and Semantic Segmentation for Autonomous Driving: Datasets, Methods, and Challenges
• A survey of synthetic data augmentation methods in computer vision
• Synthetic Datasets for Autonomous Driving: A Survey
• How much real data do we actually need: Analyzing object detection performance using synthetic and real data
• Unsupervised Intra-domain Adaptation for Semantic Segmentation through Self-Supervision
• Cityscapes 3D: Dataset and Benchmark for 9 DoF Vehicle Detection
• Learning Texture Invariant Representation for Domain Adaptation of Semantic Segmentation
• AuxNet: Auxiliary tasks enhanced Semantic Segmentation for Automated Driving
• Birds of A Feather Flock Together: Category-Divergence Guidance for Domain Adaptive Segmentation
• Adapting Object Detectors with Conditional Domain Normalization
• Ad-datasets: a meta-collection of data sets for autonomous driving
• All for One, and One for All: UrbanSyn Dataset, the third Musketeer of Synthetic Driving Scenes
• Visibility Guided NMS: Efficient Boosting of Amodal Object Detection in Crowded Traffic Scenes
• Object Detector Differences when using Synthetic and Real Training Data
• Unsupervised Traffic Scene Generation with Synthetic 3D Scene Graphs
• Co-Training for Unsupervised Domain Adaptation of Semantic Segmentation Models
• Multi-Source Domain Adaptation with Collaborative Learning for Semantic Segmentation
• Single-Shot 3D Detection of Vehicles from Monocular RGB Images via Geometry Constrained Keypoints in Real-Time
• Semi-supervised Domain Adaptation based on Dual-level Domain Mixing for Semantic Segmentation
• Scalable Modular Synthetic Data Generation for Advancing Aerial Autonomy
• MBAPose: Mask and Bounding-Box Aware Pose Estimation of Surgical Instruments with Photorealistic Domain Randomization
• SPCL: A New Framework for Domain Adaptive Semantic Segmentation via Semantic Prototype-based Contrastive Learning
• Neural Network Generalization: The impact of camera parameters
• Union-set Multi-source Model Adaptation for Semantic Segmentation
• Cross-Region Domain Adaptation for Class-level Alignment
• Self-Supervised Object-in-Gripper Segmentation from Robotic Motions
• Soft Prototyping Camera Designs for Car Detection Based on a Convolutional Neural Network
• CAR -- Cityscapes Attributes Recognition A Multi-category Attributes Dataset for Autonomous Vehicles
• Synthetic Convolutional Features for Improved Semantic Segmentation
• Predicting Depth from Semantic Segmentation using Game Engine Dataset
• An Empirical Method to Quantify the Peripheral Performance Degradation in Deep Networks
• Attention-based Adversarial Appearance Learning of Augmented Pedestrians