README.md

metadata

license: mit
task_categories:
  - token-classification
language:
  - en
pretty_name: NVR Entity Recognition Experiment
size_categories:
  - n<1K

NVR Entity Recognition Experiment

Overview

This repository contains a training dataset designed for entity recognition in Network Video Recorder (NVR) applications, specifically focused on newborn safety monitoring. The dataset uses a stuffed animal as a privacy-conscious substitute for actual newborn footage, enabling the development of computer vision models that can identify critical safety scenarios in nursery environments.

Purpose

The primary goal of this dataset is to train machine learning models capable of recognizing:

• Sleeping positions: Back sleeping (safe), side sleeping, face-down sleeping (unsafe)
• Dangerous objects: Blankets, pacifiers, and other items that could pose smothering risks
• Safety events: Various scenarios that require parental attention or intervention

Dataset Structure

Sample Images

Here are some representative examples from the dataset:

Safe Sleeping Position (Back Sleeping)

Unsafe Sleeping Position (Face Down)

Safety Event Detection (Dangerous Objects)

Camera Locations and Equipment

The dataset includes footage from multiple camera positions commonly found in home nursery setups:

• Bedroom Bassinet: Monitored using Reolink E1 Pro camera
• Living Room Buggy: Monitored using Tapo C210 camera
• Nursery Bassinet: Monitored using Tapo C200 camera

Data Categories

cam-captures/
├── bedroom-bassinet/
│   ├── back-sleeping/     # Safe sleeping position
│   ├── face-down/         # Unsafe sleeping position
│   └── side-sleeping/     # Potentially unsafe position
├── living-room-buggy/
│   ├── back-sleeping/
│   ├── face-down/
│   └── side-sleeping/
├── nursery-bassinet/
│   ├── 1/
│   ├── 2/
│   └── 3/
└── events/
    ├── blanket-in-bassinet/  # Dangerous object detection
    ├── pacifier/             # Object that could pose risks
    └── smothering/           # Critical safety scenarios

Camera Specifications

Reolink E1 Pro

• Location: Bedroom bassinet monitoring
• Features: Pan/tilt capabilities, night vision
• Use Case: Primary sleeping area surveillance

Tapo C210

• Location: Living room buggy monitoring
• Features: 360° rotation, motion detection
• Use Case: Mobile sleeping area monitoring

Tapo C200

• Location: Nursery bassinet monitoring
• Features: Fixed position, infrared night vision
• Use Case: Dedicated nursery surveillance

Applications

This dataset is intended for training models that can:

1. Automated Safety Alerts: Detect unsafe sleeping positions and alert caregivers
2. Object Recognition: Identify potentially dangerous items in sleeping areas
3. Behavioral Analysis: Monitor and analyze sleep patterns and safety compliance
4. NVR Integration: Deploy trained models directly into existing NVR systems

Disclaimer

This dataset is for research and development purposes. Any deployed safety monitoring system should be thoroughly tested and validated before use in real-world scenarios. Automated systems should supplement, not replace, direct parental supervision.