 |
- Maximizes efficiency in autonomous driving training data processing with SVDataFlow-based integrated data management system
SEOUL, South Korea, Sept. 9, 2025 /PRNewswire/ — STRADVISION, a leader in deep learning-based vision perception technology for the automotive industry, today announced the release of SVDataFlow, an advanced version of the “Data Management Workflow” first introduced at CES 2025. This upgrade, developed as part of the company’s Digital Transformation 2.0 strategy, strengthens automation and optimization capabilities for large-scale 3D training data processing and secures scalability through hybrid cloud services.
SVDataFlow: An Integrated Pipeline Connecting Data Collection to Analysis
SVDataFlow is a comprehensive data management pipeline that automates the entire process for autonomous driving and ADAS development, from sensor data collection and ingestion to preprocessing, labeling, verification, and final analysis.
Key features include:
- SURF Recording System: Synchronizes camera, LiDAR, radar, and vehicle CAN signals to ensure high-quality data.
- AI-based Time Correction: Refines sensor timestamps for improved alignment of images and point clouds.
- Auto-Labeling Engine & Auto-Sampling: Automatically labels large datasets and selects samples that directly contribute to model improvement.
- Web Labeling Tool (Labelit) + ALAS: Supports semi-automatic labeling with pre-trained labels, reducing repetitive tasks and improving efficiency.
- Quality Verification Network & WLS/WRS: Automatically validates and analyzes labeling quality, minimizing error rates and optimizing data reuse.
With this upgrade, STRADVISION has also secured scalability based on a hybrid cloud architecture that combines on-premise and cloud infrastructure. This design enables flexible responses to customer requirements: cloud resources can be expanded immediately for large-scale data processing, while projects with strict security regulations can be supported with on-premise operations to ensure compliance and stability.
Through the introduction of SVDataFlow, STRADVISION has already demonstrated significant benefits, including faster 3D data processing, a 30-40% improvement in labeling efficiency, and a noticeable reduction in operational costs. Notably, optimization of ALAS (Auto-Labeling Assistant Service) 3D map loading has significantly eased the workload of global operators.
“SVDataFlow is more than a data pipeline; it is a comprehensive management system that achieves balance across quality, productivity, efficiency, and scalability,” said Insu Kim, Head of STRADVISION’s Data Innovation Center. “We plan to commercialize SVDataFlow by the end of 2025, expand it into SVDataFlow as a Service in 2026, and roll it out globally. Through Digital Transformation 2.0, we will work with our customers to lead the global autonomous driving market.”
For more information on STRADVISION and its industry-leading technologies, please visit STRADVISION.
About STRADVISION
Founded in 2014, STRADVISION is an automotive industry pioneer in artificial intelligence-based vision perception technology for ADAS. The company is accelerating the advent of fully autonomous vehicles by making ADAS features available at a fraction of the market cost compared with competitors. STRADVISION’s SVNet is being deployed on various vehicle models in partnership with OEMs; can power ADAS and autonomous vehicles worldwide; and is serviced by over 300 employees in Seoul, San Jose, Detroit, Tokyo, Shanghai, and Dusseldorf. STRADVISION has been honored with Frost & Sullivan’s 2022 Global Technology Innovation Leadership Award, the Gold Award at the 2022 and 2021 AutoSens Awards for Best-in-Class Software for Perception Systems, and the 2020 Autonomous Vehicle Technology ACES Award in Autonomy (software category). In addition, STRADVISION and its software have achieved TISAX’s AL3 standard for information security management, as well as being certified to the ISO 9001:2015 for Quality Management Systems and ISO 26262 for Automotive Functional Safety.