JPClient: Japan
Occupancy Monitoring System / Smart office · AI analytics

Seat and area
occupancy analytics.

Occupancy Monitoring System is an AI computer vision system for smart offices. Ceiling-mounted 360° fisheye cameras detect people by their heads with a head detection model that PANTONA built and trained in-house, map each detection to a chair or area on the floor plan, and turn it into real-time and historical occupancy data: key performance indicators (KPIs), charts, heatmaps, and reports for facility and real estate teams.

360° fisheye camerasIn-house AI head modelSeat and area analytics
Illustration: a ceiling fisheye camera detects people and shows seat occupancy on a floor plan heatmap, with KPI and heatmap panels
In-househead detection model by PANTONA
4Product components
1,421QA test items passed
0Errors in 13,208 load-test requests
2.0 sAverage dashboard load at 10 users
Built in-house

Our own head detection model.

The head detection model behind Occupancy Monitoring System was built and trained by PANTONA’s own engineers. Ceiling-mounted fisheye cameras see people from directly above, an angle that general-purpose person detectors are not designed for, so we trained a model that recognizes heads from that top-down view.

01

Made for the top-down view

Trained on 360° fisheye images taken from the ceiling, where people appear as heads and shoulders instead of full bodies.

02

Built on YOLO

Uses the YOLO architecture (YOLOv9) as its base, with PANTONA’s own training and tuning for seats and areas.

03

Owned and improved by PANTONA

Because the model is ours, we can retrain and update it for new offices, layouts, and camera positions.

What it does

Main functions.

01

Register 360° cameras

Add fisheye CCTV cameras as video sources and save camera presets for each view.

02

Map chairs and areas

Draw chairs, areas, and room boundaries on the camera view and an interactive floor layout.

03

AI head detection

PANTONA’s own YOLO-based head detection model finds heads and matches them to chairs or areas, on a schedule or on demand.

04

Statistics and heatmaps

Occupancy rate, total visitors, and maximum chairs used, by date, month, hour, or weekday.

05

Integrations

Send real-time availability to room management systems such as RoomSense, and to the dashboard.

06

Export and maintenance

CSV and ZIP export, holiday and working-hour settings, logs, and module updates.

The product family

Product components.

Chair occupancy (head detection)

Seat Monitoring

Detects heads with fisheye cameras and maps them to chair locations to track real-time and historical chair use.

MySQL or MSSQLAPI to RoomSense703 tests passed
Area utilization

Area Monitoring

Measures how often user-defined areas are used, with occupancy rate, visitor count, people density, and averages.

In-house YOLOv9 modelArea grouping439 tests passed
Analytics module in the main dashboard

Seat Occupancy Dashboard Cloud

KPI cards, trends, floor-plan heatmaps, seat rankings, and peak-hour analysis from data already in the main dashboard.

Docker · nginxPeak mode158 tests passed
Local dashboard

Seat Occupancy Dashboard Stand-alone

The same seat occupancy analytics as a stand-alone application for sites without the cloud dashboard.

Stand-aloneEN / JA121 tests passed
How it works

Setup in four steps.

01
Step 1

Add a camera

Register a 360° fisheye camera as a video source and restart the application.

02
Step 2

Create presets

Save camera positions and angles for each monitoring view.

03
Step 3

Map chairs or areas

Draw chair and area shapes, group them, and place them on the room layout.

04
Step 4

Detect and analyze

Run detection by scheduler or manually, then read the statistics, heatmaps, and dashboard.

Seat Occupancy Dashboard

Dashboard KPIs.

The dashboard summarizes the selected period and space in five KPI cards, then lets users drill down by day, weekday, hour, seat, and floor.

KPI 1Average seat utilization
Average share of seat-use time over working hours
KPI 2Maximum utilization
Highest aggregated utilization in the period
KPI 3Average daily seats used
Distinct seats used per day, on average
KPI 4Operating days
Workdays counted in the selected period
KPI 5Low-utilization seats
Seats below the configured low-use threshold

Views

Daily trend (average and max)Weekday averageHourly averageDate × hour heatmapWeekday × hour heatmapFloor-plan heatmapSeat rankingPeak-hour modeChart image export
Quality assurance

Test results.

PANTONA’s QA team tests every component against detailed checksheets in Chrome and Edge, covering UI, validation, localization, features, scenarios, and statistics validation.

ComponentTest itemsPassedPass rateCoverage
Seat Monitoring703703100%Login, settings, mapping, API, statistics, statistics validation, viewer, maintenance
Area Monitoring439439100%Login, admin, area mapping, statistics, statistics validation, viewer, maintenance
Dashboard Cloud158158100%UI/UX, validation, localization, features, scenarios, rare cases, security, performance
Dashboard Stand-alone121121100%UI/UX, validation, localization, log evidence, features, scenarios, rare cases
Total1,4211,421100%

Load test: Dashboard Cloud (1, 5, and 10 concurrent users, real browsers)

Error rate0% at every load level
Average load time1.40 s · 1.58 s · 2.02 s
p95 load time at 10 users3.12 s
Comfortable capacity10–15 concurrent users (current setup)
Access control

User roles.

Super Admin

Full access: mapping, presets, maintenance, updates, holidays, and all statistics. One active session at a time.

Sub Admin

Read-only access to the viewer and statistics. Multiple logins allowed.

Viewer

Public viewer mode that shows live occupancy without changing settings.

Technical details

Technical specifications.

Operating systemWindows 10 Pro 64-bit (min.) · Windows 11 Pro or Linux Enterprise Server (rec.)
CPU12th Gen Intel Core i7 2.1 GHz (min.) · 3.6 GHz or higher (rec.)
GPU (optional)NVIDIA GTX 1650 4 GB (min.) · RTX 3060 12 GB or higher (rec.)
Memory and storage8 GB RAM, 512 GB (min.) · 16 GB RAM, 1 TB SSD (rec.)
Scale reference20 presets / 50 seats up to 50 presets / 200 seats
Supported camerasAXIS M4327-P · Hikvision DS-2CD2955G0-IS(U)

Technology

Python 3.10FlaskIn-house head detection modelYOLO (YOLOv9, Ultralytics, LibreYOLO)MySQL 8.0Microsoft SQL ServerDockernginx
Occupancy Monitoring System

Interested in Occupancy Monitoring System?

Tell us about your office, floors, and cameras, and we’ll share more details.