Reference
The classified volume count, and what a DPR needs from it
The traffic chapter of a Detailed Project Report rests on a classified volume count. This is what that deliverable has to contain, which parts can be produced from recorded video, and which still require someone standing on the road.
Typical survey periods in Indian practice
24 h
48 h
7 day
What the deliverable contains
Six things a reviewer will look for
Each of these comes out of Playcraft ATCC directly, as a file rather than as a figure someone transcribed.
Classified counts by vehicle category
All 19 canonical classes, detected per vehicle rather than sampled.
Directional split
Counted separately per direction, including opposing flows across one line on a divided carriageway.
Interval aggregation
Fifteen-minute and hourly totals across whatever period the footage covers, so peak-hour factors come straight out.
Volume in PCU
Passenger car units applied per IRC:64-1990, reported alongside the raw count rather than instead of it.
An auditable trail
Every counted vehicle logged with timestamp and confidence, so any interval reconciles back to the frame it came from.
Evidence a reviewer can check
An annotated verification video, so a disputed figure is settled by scrubbing to the timestamp rather than by argument.
What it does not replace
Three surveys video cannot do
A classified volume count is one input to a traffic chapter, not the whole of it. These need field work, and no amount of computer vision changes that.
If a vendor tells you a camera can produce an axle load spectrum, ask them how. Pavement design depends on that number being right.
Axle load survey
Needs a weighbridge or portable weigh pads in the field. Nothing in a video determines axle weight, and pavement design depends on it.
Origin–destination survey
Needs roadside interviews or number-plate matching across sites. A count at one section cannot tell you where a trip started.
Speed and delay by vehicle type over a network
We estimate speed from line-to-line crossing time at a section. That is not the same as a journey-time study.
Where the count comes from
Footage you already have
There is no roadside hardware to procure, permit or guard. CCTV already on the corridor, a drone pass, or a phone on a pole all work, and the footage never leaves your building — the software is licensed to run on your own machines.
What governs accuracy is not the camera brand but how many pixels a vehicle occupies as it crosses the counting line. We publish the method and the numbers, including where ours degrades, and every class with its PCU factor.
Before you commit a date
Send fifteen minutes from the site
We will return the classified count, the PCU summary and the annotated video, and tell you honestly what the camera position is costing you before you plan the survey around it.
Send us footage