ConductVision Enterprise
ConductVision flags the moment a person falls in camera view and times how long until someone reaches them, for senior living communities, worksites and stations.
5 detectionsConductVision detections on this photo.
Fall Detection uses pose-estimation models to identify the rapid vertical-to-horizontal body transitions characteristic of fall events, distinguishing them from normal activities like sitting down, bending, or lying on a couch. The system is designed to route an alert when a fall is detected, supporting response workflows in elderly care, assisted living, and industrial safety. Unlike wearable-based solutions, camera-based detection requires no device on the person and works for anyone within the camera field of view, visitors, staff, and residents alike.
Outcomes
A fall in a stairwell, aisle or back room may happen with nobody nearby.
The person waits until someone happens to walk past.
Work on ladders goes on all shift, and a supervisor sees only moments of it.
Slips that end on a lower rung go unreported.
Incident reports after a fall rely on what the person and any witnesses recall.
Without footage, a team cannot tell a trip over a cable from a slip on a wet floor.
What the model sees
ConductVision detections on this photo.
ConductVision detections on this photo.
Endpoints
Each endpoint is defined before the pilot starts, with what is counted, its unit and how the results are broken down.
| Endpoint | How it is counted | Unit | Broken down by |
|---|---|---|---|
| PrimaryTime to first response | Minutes from the frame where a person goes down to the first frame with another person beside them. | Unit:minutes (median) | By:Area, shift |
| SecondaryFall rate | Confirmed fall events per 1,000 hours of people in view, from the person counts in sampled frames. | Unit:falls per 1,000 person-hours | By:Area, hour |
| Ladder falls | Share of confirmed falls that start on or beside a ladder. | Unit:% | By:Area, week |
| Unanswered alerts | Fall alerts with nobody reaching the person within the window the team sets, counted per week. | Unit:alerts per week | By:Area, shift |
Each pilot measures ConductVision on your own footage against a sample checked by hand, before anyone relies on its numbers.
Fall alert precision
Share of fall alerts that a reviewer confirms as real falls in a hand-checked sample of alert clips.
% · by camera, area
Missed falls
Share of falls in a reviewed test set, including falls acted out by staff on mats, that ConductVision did not flag.
% · by camera, fall type
False alerts
Alerts a reviewer rejects, such as a person kneeling to work or stretching on a mat, per camera per day.
alerts per camera-day · by camera, activity
How it works
Connect
Install cameras in common areas, hallways, bathrooms (privacy-mode compatible), and high-risk zones. The system works with existing IP cameras and supports privacy-preserving skeleton-only processing where full video is not appropriate.
Analyze
A real-time pose-estimation model tracks body keypoints frame by frame, building a temporal sequence of posture states. A fall classifier analyzes the velocity and trajectory of the transition from upright to ground-level, filtering out normal activities like sitting or crouching.
Act
When a fall is confirmed, alerts are dispatched to designated caregivers, nursing stations, or monitoring centers with the camera location, a snapshot or skeleton-only image, and the timestamp. If no response is acknowledged within a configurable window, the alert escalates.
What it detects
Person
Each person in view, tracked with one ID
Body pose
Keypoints for the head, shoulders, hips and knees, frame by frame
Fall
The frame a person goes down, from the fall class and the pose
Ladder
Located in the scene, so falls from or beside it are tagged
Arrival
The first frame another person reaches the one who fell
Features
Use cases
Residents in assisted living may be alone when a fall occurs. Camera-based detection can notify on-duty staff so a fall does not wait for the next scheduled check.
healthcaresenior living
Hospitalized patients, especially those recovering from surgery or on sedating medications, are at elevated fall risk. Room-mounted cameras in skeleton-only mode can alert nursing staff without compromising patient privacy.
Older adults living independently face fall risk without on-site staff. A camera system in key rooms can alert family members or remote monitoring services, enabling aging in place with an additional safety layer.
healthcaresenior living
Workers on elevated platforms, ladders, and scaffolding face fall hazards. Camera-based detection can provide an alert independent of whether the worker is wearing a personal fall-arrest system, covering visitors and contractors as well.
manufacturingconstructionenergy
Customer falls on wet floors or cluttered aisles call for a fast staff response. Detection can prompt that response and generate timestamped incident documentation for risk management.
retailhospitality
Falls on escalators, platforms, and stairways at transit stations require rapid response. Automated detection can alert station personnel immediately, supplementing the coverage that human monitoring of dozens of camera feeds cannot maintain.
transportationsmart city
Technical overview
Accuracy and speed are measured on your footage and cameras during the pilot, and reported with the conditions they were measured under.
Industries
Related
Counts and tracks people in each camera view, by zone and hour.
Tracks posture, waiting and movement from pose keypoints in public spaces.
Counts dense crowds and maps where people gather, section by section.
Pilot
A pilot takes one prioritized outcome, runs on your cameras, reports the endpoints above and ends with a results review and a decision.
