STRADVISION, a leader in computer vision technology, has announced the development of a new data pipeline automation system designed to accelerate digital transformation. STRADVISION, 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 analyzable 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: the Data Preprocessing Pipeline, the Data Quality Pipeline, and the Data Efficiency Pipeline. This system operates on parallel processing cluster servers, efficiently securing high-quality data to optimize SVNet’s performance. The data is centralized in STRADVISION’s proprietary data center, a custom-built data warehouse that ensures consistent data access for all company members via a unified interface.
Also Read: Leopard & Sony Launch GS500 AI Imaging solution
The implementation of the data pipeline automation system has greatly minimized 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: This stage converts raw vehicle data into a format suitable for training. STRADVISION has addressed common issues such as data synchronization and skew, which arise from sensor data latency and skew. The company developed advanced data correction technology, significantly improving data quality and reducing manual labor.
- Data Quality Pipeline: In the past, increasing data quantity was the primary method for improving 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 maximize AI model performance, minimizing noise and improving recognition accuracy.
- Data Efficiency Pipeline: 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 optimize this process, ensuring consistent quality and reduced labor costs. The AI model continuously retrains and updates in real-time, supporting and streamlining the work of data generation employees.
Jack Sim, CTO of 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 optimizing resource utilization 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.
Source: PRNewswire
