Infant Monitoring Environment Image Segmentation System Dataset Overview

Dataset Information

Dataset Name

• Name: baby-monitoring

Dataset Categories

• Number of Classes: 4
• Class Names: [baby, knife, scissors, socket]

Dataset Description

• Purpose: Enhance the safety of infant monitoring environments by accurately identifying potential hazardous objects, ensuring infant safety in daily life.
• Class Selection: Based on a comprehensive analysis of possible danger factors in infants’ everyday surroundings.
• Sample Diversity: Emphasis on diverse and representative samples to ensure model effectiveness in real applications.
• Annotation Process: Professional annotation tools were used to precisely segment each image, ensuring accurate bounding of objects for each class.
• Dataset Scale: Contains several thousand images covering a wide range of scenes and objects.

Dataset Construction

• Baby Class: Includes images of babies in various poses and expressions, covering different lighting conditions and backgrounds.
• Knife Class: Collects images of various types and styles of knives and scissors, including kitchen knives and child‑safe scissors.
• Scissors Class: (Content omitted in original; retained as is.)
• Socket Class: Socket images were captured from multiple angles and environments to improve recognition accuracy.

Dataset Applications

• Goal: Develop an efficient and accurate image segmentation system to provide robust technical support for infant safety monitoring, thereby advancing the development and application of infant monitoring technologies.