Machine learning assisted annotation leveraging large data foundation models

• Main Author: Tey, Chin Yi
• Other Authors: Lin Guosheng
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Machine learning; Data annotation; Semantic segmentation; Computer vision; Deep learning models
• Online Access: https://hdl.handle.net/10356/175707
• Institution: Nanyang Technological University

This research paper explores the potential of machine learning-assisted annotation methods in streamlining the data annotation process, enhancing annotation quality, and unlocking the full potential of annotated datasets at a production level. The author investigates the use of various advanced machine learning models, including the Self-supervised Equivariant Attention Mechanism for Weakly Supervised Semantic Segmentation (SEAM), Sobel Image Edge Detection, and the Segment Anything Model (SAM), to address challenges in large-scale data annotation. The study demonstrates that machine learning-assisted annotation methods represent a crucial solution to bridge the gaps in SAM, forming the basis for future ML-assisted annotation. The research highlights the significance of continued innovation in machine learning and computer vision to advance state-of-the-art data annotation practices and methodologies.