How does computer vision occupancy detection differ from traditional PIR motion sensors?

Traditional PIR (Passive Infrared) motion sensors detect the presence or absence of movement within their field of view but cannot count how many people are present, distinguish between individuals, or detect stationary occupants (someone sitting still at a desk reading, for example, may not register continuous motion and can cause a PIR sensor to falsely indicate an unoccupied space). Computer vision occupancy detection uses camera-based person-detection AI models to count actual occupant numbers within a zone in real time, correctly identifying stationary occupants, providing occupancy density data (not just binary occupied/unoccupied), and enabling more granular zone-based control decisions. The tradeoff is that computer vision requires camera installation and more sophisticated data processing than a simple PIR sensor, along with more careful privacy architecture design given the camera-based data source.

Does occupancy-based automation compromise employee privacy under India's DPDP Act 2023?

Occupancy-based automation can be designed to avoid DPDP Act 2023 personal data classification by using anonymised body-count analytics rather than identity-linked detection. Computer vision occupancy counting models are typically configured to output only aggregate person counts per zone (e.g., 'Zone 4: 12 people present') without performing facial recognition or any individual identification — the video frame is processed locally at the edge, and only the anonymised count data (not the raw video or any biometric template) is transmitted to the BMS. This anonymised architecture means occupancy-based automation typically falls outside DPDP's sensitive personal data provisions, since no personal or biometric data is being processed or retained. ASDV recommends this privacy-by-design anonymised architecture as standard practice for all occupancy-based automation deployments in India.

How much energy can occupancy-based HVAC control save in a hybrid-work office?

Occupancy-based HVAC automation delivers 18-25% energy savings in variable-occupancy commercial spaces, based on Indian hybrid-work office deployment studies, by conditioning only zones with actual detected occupancy rather than following fixed schedules that assume full building use. The savings magnitude correlates directly with occupancy variability — buildings with highly predictable, consistent occupancy (traditional 9-to-6, full-attendance offices) see lower savings from occupancy-based control since actual occupancy already closely matches scheduled assumptions, while buildings with genuinely variable, unpredictable daily attendance (typical of post-pandemic hybrid-work Indian corporate offices operating at 40-60% average occupancy) see savings toward the higher end of the range, since fixed-schedule conditioning was previously wasting substantial energy conditioning consistently under-occupied space.

Can occupancy-based automation predict when people will arrive, not just react to their presence?

Yes — advanced occupancy-based automation systems combine real-time detection with predictive occupancy modelling that incorporates calendar/meeting room booking data, historical attendance patterns, and building access system data (turnstile entry trends) to pre-condition zones before occupants physically arrive. This addresses the primary limitation of pure reactive occupancy detection: if a zone only begins conditioning once occupancy is detected, there is an unavoidable comfort recovery lag (typically 10-20 minutes for HVAC to bring a zone to setpoint) during which arriving occupants experience discomfort. Predictive pre-conditioning uses booking data to begin conditioning a meeting room 15-20 minutes before a scheduled meeting starts, and uses historical arrival pattern data to begin conditioning general office zones ahead of the typical morning arrival wave, eliminating the comfort lag while still avoiding the energy waste of conditioning genuinely unoccupied zones.

What is the response time between occupancy detection and HVAC/lighting adjustment?

Lighting response to occupancy detection is typically near-instantaneous (under 1 second) since lighting systems have minimal thermal lag and can respond immediately to a detected occupancy change. HVAC response is inherently slower due to the thermal mass and mechanical lag of air conditioning systems — a zone transitioning from unoccupied setback to occupied comfort conditioning typically takes 10-20 minutes to reach full setpoint recovery, depending on the magnitude of the setback and the zone's cooling/heating capacity relative to its volume. This differential response time is why predictive pre-conditioning (anticipating occupancy before it occurs) is valuable specifically for HVAC control, while pure reactive control is generally sufficient for lighting given its near-instant response capability.

How does occupancy-based automation handle hot-desking and flexible seating arrangements?

Hot-desking and flexible seating environments require finer-grained occupancy detection than traditional fixed-assignment office zones, since the specific desks or areas occupied on any given day vary unpredictably. Effective approaches include: (1) Desk-level sensors — individual desk occupancy sensors (often integrated into desk booking systems) providing granular data that the BMS aggregates into zone-level HVAC/lighting control decisions; (2) Computer vision zone density mapping — cameras covering open-plan flexible seating areas providing real-time density heat-maps that inform HVAC zoning decisions dynamically rather than following fixed zone assumptions; (3) Desk booking system integration — where hot-desk reservation systems are in use, booking data provides advance notice of expected occupancy distribution, enabling predictive pre-conditioning of the specific area where desks have been booked for that day, even without waiting for physical occupancy detection.

Is occupancy-based automation accurate enough to avoid discomfort from delayed conditioning?

Well-designed occupancy-based automation systems avoid discomfort through a combination of conservative setback limits (unoccupied zones are set back to an energy-saving but not extreme temperature, typically 2-4°C from occupied setpoint rather than fully off, reducing the recovery time required) and predictive pre-conditioning (using calendar and historical pattern data to begin conditioning ahead of anticipated arrival rather than waiting for physical detection). Poorly designed systems that use aggressive full setback combined with purely reactive detection can create noticeable comfort lag and occupant frustration — this is a common failure mode in early or poorly commissioned occupancy-based automation deployments. ASDV designs occupancy-based automation with conservative setback parameters and predictive pre-conditioning as standard practice specifically to avoid this comfort risk.

What sensors and cameras are required for computer vision occupancy detection in an existing building?

Computer vision occupancy detection typically requires: ceiling-mounted or corner-mounted cameras with a field of view covering the zone being monitored (existing CCTV cameras can sometimes be repurposed if resolution and positioning are adequate, though dedicated occupancy analytics cameras with optimised field-of-view and processing are more common for purpose-built deployments); an edge processing unit (either integrated into the camera or a nearby gateway device) running the person-detection AI model locally to generate anonymised count data without transmitting raw video; network connectivity (PoE for camera power and data, or wireless for retrofit scenarios) to the BMS/IoT gateway; and integration software mapping camera zones to corresponding BMS HVAC/lighting control zones. For a typical open-plan office floor, camera density is approximately 1 camera per 150-250 sqm depending on ceiling height and desired detection granularity.

Occupancy-Based Automation: Computer Vision HVAC & Lighting Control for Real-Time Building Use

A fixed-schedule BMS conditions the entire office floor from 8am to 7pm every weekday, on the assumption that the floor will be fully occupied throughout those hours. In the post-pandemic hybrid-work era, that assumption is simply false — Indian corporate offices average 40-60% occupancy on any given day, with substantial day-to-day and zone-to-zone variability. The BMS is conditioning empty desks, empty meeting rooms, and empty zones at full capacity, burning energy for occupants who are working from home that day.

Occupancy-based automation replaces the fixed-schedule assumption with real observation. Computer vision occupancy detection counts actual people present in each zone in real time — not motion, not a schedule, but genuine occupancy density — and the BMS conditions accordingly. Combined with predictive pre-conditioning informed by meeting room bookings and historical attendance patterns, the building delivers precision comfort exactly where and when it's needed, while eliminating the energy waste of conditioning empty space that fixed schedules cannot avoid.

Computer vision occupancy-based HVAC and lighting automation reduces energy consumption in variable-occupancy commercial spaces by 18–25% — by conditioning only zones with actual detected occupancy rather than following fixed schedules that assume full building use. CBRE / occupancy analytics deployment study for Indian hybrid-work offices, 2025.

Occupancy Detection Method Comparison

Detection Method	Accuracy	Zone Granularity	Response Time	Privacy Considerations	Energy Savings Potential
Fixed schedule (baseline)	N/A — no detection	None (assumed full)	N/A	None	0% (reference)
PIR motion sensors	Binary (occ/unocc)	Zone-level	Instant (motion only)	Low risk	10–15%
Computer vision counting	High — actual count	Sub-zone/desk-level	Real-time	Anonymised architecture required	18–25%
Wi-Fi/BLE device detection	Medium — device proxy	Zone-level	Real-time	Device ID anonymisation needed	12–18%
Desk sensor + badge correlation	High — direct + validated	Desk-level	Real-time	Badge data is personal data	15–22%

Technical Design: Occupancy-Based Automation Architecture

Computer vision occupancy counting: Anonymised person-detection models (not face recognition) provide real-time zone occupancy count without identity capture — processed at the edge, only aggregate count data transmitted to BMS
Zone-based HVAC modulation: VAV damper and AHU response to occupancy-derived cooling load, avoiding conditioning of unoccupied floors/zones while maintaining comfort in actively used areas
Lighting integration: DALI/KNX lighting control responding to the same occupancy data feed as HVAC, delivering coordinated automation across both systems from a unified data source
Hybrid work pattern adaptation: BMS control logic redesigned around genuinely variable, unpredictable daily occupancy rather than fixed 9-to-6 assumptions — reflecting India's 40-60% average post-pandemic office occupancy reality
Privacy-preserving architecture: DPDP Act 2023-conscious design using anonymised body-count analytics — avoiding biometric/identity data classification through edge processing and aggregate-only data transmission
Predictive occupancy modelling: Meeting room booking data and historical attendance patterns combined with real-time detection to pre-condition zones 15-20 minutes ahead of anticipated arrival, eliminating comfort recovery lag
Access control cross-reference: Turnstile entry data as a secondary occupancy signal source, cross-referenced with computer vision counting for accuracy validation and redundancy
Hot-desking granularity: Desk-level sensors or booking system integration providing finer-grained occupancy data for flexible seating environments where zone-level assumptions break down

Future Outlook: 2028–2035

Individual Comfort Profiles: Occupancy Meets Personalisation

Future occupancy-based automation will move beyond zone-level aggregate response toward individual comfort personalisation — combining anonymised occupancy detection with opt-in personal comfort preference profiles (accessed via a mobile app, not tied to identity in the BMS) to deliver micro-zone conditioning tuned to the specific mix of people present in a space at any moment. A meeting room's HVAC will automatically bias toward the average preference of the specific attendees who have checked in for that meeting, without the BMS ever needing to know who those individuals are — preference data remaining under the individual's control on their device, contributed anonymously and aggregated only at the point of use.

Frequently Asked Questions

PIR sensors detect presence/absence of movement but cannot count occupants or reliably detect stationary people (someone sitting still reading may falsely register as unoccupied). Computer vision uses camera-based person-detection AI to count actual occupant numbers in real time, correctly identifying stationary occupants and providing density data for more granular zone control. The tradeoff is camera installation requirements and more careful privacy architecture design.

It can be designed to avoid DPDP personal data classification through anonymised body-count analytics — computer vision models output only aggregate person counts per zone without facial recognition or individual identification, processed locally at the edge with only count data (not raw video or biometric templates) transmitted to the BMS. This anonymised architecture typically falls outside DPDP's sensitive personal data provisions. ASDV recommends this privacy-by-design approach as standard practice.

Yes — advanced systems combine real-time detection with predictive modelling using calendar/meeting room booking data and historical attendance patterns to pre-condition zones before physical arrival. This addresses the comfort recovery lag inherent to pure reactive detection (HVAC typically takes 10-20 minutes to reach full setpoint), using booking data to begin conditioning meeting rooms 15-20 minutes before scheduled start and historical patterns to anticipate morning arrival waves.

Occupancy-Based Automation: Computer Vision HVAC & Lighting Control for Real-Time Building Use

Occupancy Detection Method Comparison

Technical Design: Occupancy-Based Automation Architecture

Occupancy-Based Automation Design

Individual Comfort Profiles: Occupancy Meets Personalisation

Frequently Asked Questions

Design Your Occupancy-Based Automation System