Applications / Operations Intelligence
Line Status & Stoppage Detection
Detects whether a line is running or visually stopped, times every stoppage and micro-stop, and captures the reason on video as it happens.
See it in action
Twenty-two minutes of micro-stops nobody logged
Video slot — awaiting upload.
What it detects
Downtime that is logged gets fixed. Micro-stops, short stalls and unlogged interventions are the ones that quietly take a shift apart. Fogsphere sees all of them the same way.
Line running or visually stopped
Classifies each line as running or stopped from the visual state of product flow, without a PLC tap.
Micro-stop frequency and duration
Counts short stoppages that fall below the threshold operators bother to log, and reports mean stop duration.
Machine idle and running state
Detects machine idle versus running state, and logs each transition with a clip.
Blocked and starved conveyors
Detects conveyor congestion, empty conveyor, and lines starved or blocked upstream and downstream.
Downtime reason video
Attaches the seconds either side of a stop to the stop record, so the reason is reviewable rather than guessed.
Andon and callpoint confirmation
Confirms a line stop visually, detects worker interaction with the callpoint, and times supervisor response.
Manual intervention counts
Counts manual interventions, and how long the line ran degraded before or after each one.
Product accumulation and reject build-up
Detects product accumulating ahead of a stop, and rejects building up at a reject station.
Robot and cell stoppage
Detects blocked robot paths, stopped cells and AGV dwell inside automated areas.
Operator absence at required stations
Flags a station that should be staffed while the line is running, and is not.
Restart and first-production timing
Times the gap between a stop clearing and the first unit produced after restart.
Where it applies
Any process with a running state and a stopped state can be classified visually. The Q3 papers describe it on production lines, packaging halls, conveyors and sortation.
How it works
Edge inference on the cameras you already have
Line-state classification runs on the edge device next to the camera already pointing at the line. Lines with no PLC signal, no OEE instrumentation and no reason-code discipline are covered the same way as new ones.
Zones, thresholds and schedules you configure
Draw the line segment or cell, set how long a stall must last before it counts as a stop, and set the shift pattern availability is measured against. Each segment can carry its own thresholds.
Every event carries its own evidence
Each detection is logged with a timestamp, the camera and zone it came from, and a short clip. The record can be reviewed, exported or attached to an investigation without trawling through hours of footage.
Feeds the systems already in place
Events surface in Fogsphere Manager and route outward to the MES, OEE dashboard, andon board, maintenance queue or messaging channel operators already watch.
Risk & control objective
The control objective is availability that reflects what happened, not what was typed into a reason code. Long stops are logged reliably; short ones are absorbed, and a shift can lose an hour in ninety-second pieces nobody recorded.
The value is the explanation attached to the number. OEE reports that availability fell. The visual record shows six product accumulations, three manual interventions and twenty-two minutes with no operator at the station — which is the difference between a metric and a task list.
Over weeks the same detections turn stoppage into a pattern: the same machine at the same point in the shift, or the same changeover always followed by a slow restart. That is where the recoverable hours sit.
Related solutions
See every stop, including the ones nobody logs
We will show you line-state classification, micro-stop counting and downtime reason capture on your own camera estate.