ADQ is an optimized tool for data quality verification that utilizes AIWORKX' experience and know-how in large-scale AI data construction technology.
Experience in Constructing and Verifying AI Data
-
Experience in constructing large-scale data
-
Latest AI automation technology
-
Experience in a variety of quality verification techniques
-
KOLAS-Accredited Institution
Quality Verification Tool Optimized for Data Verification
ADQ(AI Data Quality) is an AI data quality verification tool. It was developed by utilizing the in-house experience and capability of constructing and verifying large-scale AI training data in various domains.
- Image data verification
- Video data verification
Effectiveness of Quality Verification through ADQ
- Increase efficiency and productivity
- Make a foundation for constructing high-quality data sets through error analysis
4 Steps of ADQ Quality Verification
- Convert files to load result filesFormats supported for conversion: CVAT xml, Pascal VOC xml, COCO json
- Create project and task, after sampling the verification data
- Conduct verification
- Check verification result
Output numerical data and the report
Main Features
-
Project management
- Multiple projects are possible
- Efficient status management by project
- Creation of multiple verification tasks within one project
-
Sampling auto-creation
- Support conversion of various formats
- Specify the desired sample quantity
- Support auto-creation of sampling data and tasks
-
Statistics
- Supports accuracy verification of semantics and syntax
- Supports statistical dashboard for each project according to the verification progress
-
Inspection and analysis
- Supports inspection to compare GT values and result files (using IOU)
- Supports data analysis
(label and object size distribution)
The quality of training data is very important for the development of trustworthy AI services. In particular, considering the huge cost of data construction, it would be ideal to secure the minimum amount of high quality data for successful AI data construction.
In order to secure high-quality data, continuous quality management is required from the beginning. But in reality, domestic AI data quality verification is still in its infancy. In addition to this, most of the data-related tools used in AI data building businesses are focused on data labeling, auto-labeling, and labeler task management.
ADQ was developed according to the accuracy quality index of data labeling by reflecting the difficulties that the companies and institutions in this field face while performing AI data quality verification. Through ADQ, companies or institutions that need quality verification guides or third-party quality verification can establish a relevant process, manage the verification, and check the resulting analysis report post verification. AIWORKX' quality verification tool is expected to lead to the expansion of the scope of data quality verification which is imperitive in determining the quality of AI services.
AIWORKX Founder& CEO Dale Yoon