How much real data do we actually need: Analyzing object detection performance using synthetic and real data

Abstract

In recent years, deep learning models have resulted in a huge amount of progress in various areas, including computer vision. By nature, the supervised training of deep models requires a large amount of data to be available. This ideal case is usually not tractable as the data annotation is a tremendously exhausting and costly task to perform. An alternative is to use synthetic data. In this paper, we take a comprehensive look into the effects of replacing real data with synthetic data. We further analyze the effects of having a limited amount of real data. We use multiple synthetic and real datasets along with a simulation tool to create large amounts of cheaply annotated synthetic data. We analyze the domain similarity of each of these datasets. We provide insights about designing a methodological procedure for training deep networks using these datasets.

Accepted in International Conference on Machine Learning (ICML 2019) Workshop on AI for Autonomous Driving

References in corpus (11)

• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• End to End Learning for Self-Driving Cars
• ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation
• FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation
• Virtual Worlds as Proxy for Multi-Object Tracking Analysis
• Dynamic Graph CNN for Learning on Point Clouds
• What Makes Good Synthetic Training Data for Learning Disparity and Optical Flow Estimation?
• Augmented LiDAR Simulator for Autonomous Driving
• Synscapes: A Photorealistic Synthetic Dataset for Street Scene Parsing
• Complex-YOLO: Real-time 3D Object Detection on Point Clouds
• Procedural Modeling and Physically Based Rendering for Synthetic Data Generation in Automotive Applications

Cited by in corpus (7)

• Radar-Camera Fusion for Object Detection and Semantic Segmentation in Autonomous Driving: A Comprehensive Review
• StorSeismic: A new paradigm in deep learning for seismic processing
• Quantifying the LiDAR Sim-to-Real Domain Shift: A Detailed Investigation Using Object Detectors and Analyzing Point Clouds at Target-Level
• Object Detector Differences when using Synthetic and Real Training Data
• VersatileGait: A Large-Scale Synthetic Gait Dataset with Fine-GrainedAttributes and Complicated Scenarios
• Sim2Air - Synthetic aerial dataset for UAV monitoring
• Development of Hybrid Artificial Intelligence Training on Real and Synthetic Data: Benchmark on Two Mixed Training Strategies