STRADVISION Develops New Data Pipeline Automation System to Accelerate Digital Transformation
Delivering Fast, Stable, and Best-in-Class Solutions to Customers
STRADVISION, a leader in computer vision technology, has announced the development of a new data pipeline automation system designed to accelerate digital transformation. The company, known for its AI-based image recognition software, SVNet, aims to enhance its offerings by streamlining its data processes.
A data pipeline involves the preprocessing of raw data collected from various sources, converting it into an analysable format, and storing it in a data warehouse. Traditionally, this process includes Extracting, Transforming, and Loading (ETL) data. However, due to the complexity of SVNet’s requirements for precise data, STRADVISION’s ETL process is divided into several stages, making the entire workflow intricate and time-consuming.
To address these challenges, STRADVISION developed an automation system comprising three primary components: Data Preprocessing Pipeline, Data Quality Pipeline, and Data Efficiency Pipeline. This system operates on parallel processing cluster servers, efficiently securing high-quality data to optimize SVNet’s performance. The data is centralised in the company’s proprietary data centre, a custom-built data warehouse that ensures consistent data access for all company members via a unified interface.
Customers Are Already Benefiting from STRADVISIONās Innovation
The implementation of the data pipeline automation system has greatly minimised errors, boosted productivity, and significantly reduced costs. As a result, customers now enjoy faster, more accurate, and highly reliable SVNet services.
- Data Pre-Processing Pipeline.Ā This stage converts raw vehicle data into a format suitable for training. STRADVISION has addressed common issues such as data synchronisation and skew, which arise from sensor data latency and skew. The company developed advanced data correction technology, significantly improving data quality and reducing manual labour.
- Data Quality Pipeline.Ā In the past, increasing data quantity was the primary method for improving Artificial Intelligence (AI) learning outcomes. However, STRADVISION focuses on data quality, which has proven to enhance image recognition rates. The data quality pipeline employs error identification and correction to maximise AI model performance, minimising noise and improving recognition accuracy.
- Data Efficiency Pipeline.Ā This pipeline reduces the costs associated with data generation for SVNet training. Traditionally, manual data generation work varied in quality and cost. STRADVISION developed an AI model to automate and optimise this process, ensuring consistent quality and reduced labour costs. The AI model continuously retrains and updates in real-time, supporting and streamlining the work of data generation employees.
Jack Sim, CTO at STRADVISION, commented: “We are accelerating digital transformation to deliver fast and stable services to our customers. Our new data pipeline automation system is a key component of this initiative. By reducing operational costs and optimising resource utilisation through continuous process automation, we aim to provide even better services.”
As STRADVISION prepares for an initial public offering (IPO) on the KOSDAQ market later this year, the company has been intensifying its digital transformation efforts since April. It has already passed the technology evaluation for the KOSDAQ technology special listing with high marks, particularly in technology and marketability. For more information on STRADVISION and its cutting-edge technologies, please visitĀ STRADVISION.