Technical articles · 2023.09
Quality Verification of Supervised-Learning Label Data: A Focus on Computer Vision
지도학습 라벨데이터 품질 검증 방법 - 컴퓨터 비전 분야 중심
Overview
Explains checks for statistical sufficiency and balance, annotation suitability, syntactic and semantic accuracy, and validity of supervised-learning label data. It outlines a framework for choosing evaluation metrics for computer-vision tasks such as classification, detection, segmentation and pose estimation.
Publication and version information
Technical article in TTA Journal 209 (September/October 2023), pp. 59–64. It covers label-data quality verification across computer vision and refers to TTAK.KO-10.1339 and TTAK.KO-10.1420.
Author-supplied source information
Author-supplied information on arXiv.
TTA Journal209, Sept/Oct2023, pp59–64. Technical-method article; citedTTAK.KO-10.1339 andTTAK.KO-10.1420 are separate standards.
Version: TTA Journal 209