2024 10 Eigen User Forum 8 Session 5 Image recognition
Автор: Eigen
Загружено: 2024-10-18
Просмотров: 35
Discussed using machine learning and image recognition to digitize P and IDs, which are often stranded on paper due to their complex shapes.
Speaker 2 demonstrated a streamlit front end that uses the Yolo V5 object detection algorithm to identify instruments in P and ID images, achieving 98% detection accuracy with 1200 self-annotated data snippets.
The system also includes optical character recognition (OCR) to read and clean instrument text, and a naming convention builder for customization. The process involves adjusting grid size and confidence thresholds to optimize detection accuracy, and outputs a CSV file for further use.
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке: