From Recording Camera to Care System
Traditional CCTV cameras record footage for post-incident review. AI video analytics transforms cameras into proactive care systems that analyze behavior in real-time and alert caregivers before situations become emergencies.
For families caring for elderly relatives remotely, this technology represents a fundamental shift in the quality of remote care.
AI Capabilities Available in 2025
1. Fall Detection AI analyzes body pose estimation and detects rapid changes in height — the signature of a fall.
- 85-92% accuracy in testing conditions
- False positive rate approximately 8-12% (rapid sitting can trigger)
- Response time < 5 seconds post-fall
Solutions:
- Aqara FP2 (4,000-5,000 THB) — mmWave radar, not camera (better privacy)
- Nest Cam with Google Home — Fall Detection via Google Home (Beta in Thailand)
- YOLOv8 Custom Model + IP Camera + Home Assistant — DIY highest accuracy, requires technical skill
2. Activity Pattern Analysis
AI learns the individual’s daily normal: wake time, kitchen visits, outings, and time spent in each area.
If the person emerges from the bedroom 2 hours late → alert. If no kitchen activity at the usual meal time → alert.
Tools:
- Home Assistant Person Tracking + motion sensor network
- Roboflow + IP camera for custom activity detection
3. Prolonged Stillness Detection
If the elderly person remains motionless at the same location beyond a threshold (e.g., 30 minutes in the bathroom), the system alerts immediately:
yaml
alias: "Elderly Stillness Alert - Bathroom"
trigger:
- platform: state
entity_id: binary_sensor.bathroom_motion
to: "off"
for:
minutes: 30
condition:
- condition: state
entity_id: binary_sensor.bathroom_door
state: "on"
action:
- service: notify.mobile_app_family_line
data:
title: "Check Bathroom Urgently!"
message: "Elderly family member in bathroom for 30+ minutes with no movement detected."
data:
push:
sound:
name: default
critical: 1
4. Wandering Detection
For elderly with confusion or early dementia:
- Define safe zones in Home Assistant
- If leaving home after 22:00 → immediate alert
- If front door opens at night → auto-lock + critical alert
yaml
alias: "Night Wandering Alert"
trigger:
- platform: state
entity_id: binary_sensor.front_door
to: "on"
condition:
- condition: time
after: "22:00:00"
before: "06:00:00"
action:
- service: lock.lock
target:
entity_id: lock.front_door
- service: notify.mobile_app_family_line
data:
title: "Urgent Alert!"
message: "Front door opened at night. Auto-locked. Please check."
data:
push:
sound:
name: default
critical: 1
Privacy: Balancing Safety and Dignity
Thai elderly place high value on privacy and dignity. Camera-based monitoring requires explicit consent and transparent process.
Camera-Free AI Monitoring Alternatives:
- mmWave Radar (Aqara FP2) — fall and presence detection without images
- PIR Motion Sensor Network — activity pattern without facial recognition
- Smart Mattress/Chair Sensor — sit/stand detection without any camera
3-Tier AI Elderly Monitoring Packages
Basic (15,000-25,000 THB)
- mmWave radar x3 (bedroom, bathroom, living room)
- PIR motion sensor x5
- Activity pattern dashboard
- LINE alerts to family caregiver
Standard (35,000-55,000 THB)
- Everything in Basic, plus:
- 4K IP camera in common areas
- Fall detection (Aqara FP2 x2)
- Wandering detection + smart lock
- Weekly activity report
Premium (65,000-100,000 THB)
- Everything in Standard, plus:
- Custom AI model for individual activity recognition
- 24/7 cloud monitoring service
- Daily video AI analysis
- Monthly family report with trend analysis
Conclusion
AI video analytics for elderly care is not surveillance — it is intelligent, proactive caregiving. It enables longer Aging in Place while giving working adult children the confidence that their parents are monitored around the clock by systems that respond in seconds, not minutes.