AVCC vs ATCC: Sensor-Based vs Video-Based Vehicle Classification — Which One Your Toll Plaza Actually Needs

Ask any toll plaza manager what causes the most friction at the barrier, and classification disputes will be near the top of the list. A driver insists the vehicle is a car. The system has charged it as a light commercial vehicle. The lane backs up while someone checks the footage.

Multiply that by a few hundred incidents a month and you are not looking at a technology problem any more. You are looking at throughput, staffing cost, and a revenue figure nobody fully trusts.

Both AVCC and ATCC exist to solve this. They solve it differently, and the difference matters when you are writing a specification or evaluating a bid.

The short answer

AVCC (Automatic Vehicle Classification and Counting) uses physical sensing — 2D/3D LiDAR, laser profiling, and edge computing to measure a vehicle’s physical properties: axle count, height, length, and spacing. It classifies based on what the vehicle physically is.

ATCC (Automatic Traffic Counter and Classifier) uses AI video analytics. Cameras watch the traffic stream, and deep learning models recognize vehicle categories from their visual appearance, while also reporting speed, occupancy and flow.

One measures. The other recognizes. That single distinction drives almost every practical difference between them.

How AVCC works

An AVCC installation sits at the lane or gantry and builds a physical profile of each passing vehicle. Laser and LiDAR sensors capture the vehicle’s silhouette and dimensions; axle detection establishes the axle count and spacing. Edge processing turns that profile into a class in milliseconds, and passes it to the toll management system for transaction validation.

Because the classification is derived from measurable physical attributes rather than appearance, it holds up well in the conditions that break camera-only systems — heavy rain, fog, dust, glare, night operation. It is also the natural fit where classification directly determines the amount charged, because the output is defensible: you can show why a vehicle was placed in a class.

Modern AVCC systems reach up to 100% counting accuracy and 98%+ classification accuracy under normal operating conditions, and are built for continuous 24×7 operation with remote diagnostics.

How ATCC works

ATCC takes a different route. A camera stream feeds an AI model that detects and classifies vehicles two-wheelers, three-wheelers, cars, light motor vehicles, trucks and buses, multi-axle vehicles without loops, piezo sensors or in-pavement hardware of any kind.

That “no extra hardware” characteristic is the whole point. There is nothing embedded in the road surface, which means no pavement cutting, no lane closure for installation, and far less to maintain. A unit can be mounted, calibrated and moved. For traffic surveys, corridor studies and locations where you need data rather than a billed transaction, that portability is worth a great deal.

An AI-based ATCC will also give you what a physical classifier typically will not: average speed, occupancy, traffic flow patterns, pedestrian counting, and incident detection from the same video feed.

Side-by-side

Choosing between them

Rather than asking which is better, ask what the output is being used for.

If classification decides the toll amount, lead with AVCC. Anything that feeds a billed transaction barrier lanes, Open Road Tolling, Multi-Lane Free Flow corridors — needs a classification you can audit and defend when a concessionaire, an authority or a user disputes it. Physical measurement gives you that.

If classification produces data, ATCC is usually the better economics. Traffic census work, corridor planning, economic assessment, feasibility studies, before-and-after studies around a new interchange here you need volume and category distribution, not a per-vehicle billing decision. Being able to install a unit in an afternoon and relocate it next quarter changes the cost of the whole programme.

If the site is revenue-critical and operationally complex, run both. This is more common than people assume. AVCC validates the transaction; ATCC layers on speed, occupancy and flow analytics for the same corridor. The two outputs cross-check each other, which is exactly what you want when reconciliation is questioned.

What to specify in an RFP

Vendors quote headline accuracy figures. Those figures mean very little without the conditions attached. Six things worth writing into the specification:

1. State the accuracy conditions. Accuracy at what speed range, in what lighting, in what weather, across which vehicle classes. A number without conditions is not a commitment.
2. Define the class list explicitly. Indian traffic includes categories that generic models handle poorly — three-wheelers, Tata Ace-type vehicles, minibuses, OSV/EMV. List every class you expect to be distinguished.
3. Ask for the disagreement handling. When the classifier and the FASTag record disagree, what happens? Silent override is not an answer.
4. Require an audit trail. Image capture, timestamp, class assigned, transaction reference. Without this, every dispute becomes a manual investigation.
5. Specify integration, not just the device. The system must talk to your Toll Management System, ANPR, ETC/FASTag, VMS and command centre. Ask for the API, not a promise.
6. Set the acceptance test on your own traffic. Factory figures are irrelevant. Test against a manually verified sample from the actual site, across a full day-night cycle.

Why choose GreenTech ITS

Specification decisions like AVCC vs. ATCC only pay off if the vendor can actually design, build, install and support the system for the long haul. A few things are worth knowing about GreenTech ITS – Best Toll Management Service Provider before you shortlist:

• Full-stack, not just a sensor vendor. GreenTech ITS is an Original Equipment Manufacturer with in-house CAD design and testing facilities, so both AVCC (sensor-based) and ATCC/video-based vehicle profiling are built, integrated and quality-checked under one roof rather than assembled from third-party black boxes.
• Track record at scale. The company has delivered 150+ projects over 12+ years in intelligent transportation, backed by a workforce of 250+ on-site and off-site staff — the kind of bench strength a PAN-India tolling rollout or multi-corridor ATCC survey programme needs.
• Certified to the standards procurement teams ask for. STQC certification (Government of India software quality standard), CMMI Level 3 appraisal, ISO certifications covering quality management, IT service management and information security, and CE compliance are all in place — relevant when you're writing the compliance section of an RFP.
• One vendor across the whole classification stack. GreenTech's product line runs from AVCC and ANPR through AI-powered MLFF, Smart Electronic Toll Management, Weigh-in-Motion, ATMS and Video-Based Vehicle Profiling — so a "run both AVCC and ATCC" recommendation, or a later upgrade from barrier tolling to MLFF, doesn't mean a second vendor and a second integration effort.
• Support built for 24×7 tolling operations. The company operates a 24/7 command and control centre and positions its systems for continuous, energy-efficient, vandal-resistant operation — matching the always-on nature of toll and highway infrastructure rather than a business-hours support model.
• Installation and commissioning as part of the offer, not an afterthought. Supervision & Installation and Annual Maintenance Contracts are offered as standing services, alongside sustainable civil infrastructure work (toll plazas, booths, gantries, FOBs) for sites that need civil works done alongside the electronics.

Planning a classification upgrade?

GreenTech ITS designs, supplies, installs, commissions and maintains both sensor-based AVCC and AI video-based ATCC systems — with 12+ years in intelligent transportation, 150+ delivered projects, and STQC, CMMI Level 3 and ISO certification behind the delivery.

If you are writing a specification or evaluating options for a corridor, talk to our engineering team about what fits your traffic profile.

Call: 011-35456885 | Email: saurabh.khanna@greentechits.com

FAQs about AVCC and ATCC

What is the difference between AVCC and ATCC? AVCC classifies vehicles by physically measuring them using LiDAR, laser profiling and axle detection. ATCC classifies them using AI video analytics that recognise vehicle types from camera footage. AVCC is generally preferred where classification determines the toll charged; ATCC is preferred for traffic data collection, surveys and analytics.

Which is more accurate for toll collection? For revenue-critical tolling, sensor-based AVCC is normally the stronger choice because its output is based on measurable physical attributes and is easier to audit in a dispute. Well-implemented ATCC performs strongly for counting and classification but is more sensitive to camera conditions.

Can ATCC work without loops or in-pavement sensors? Yes. AI-based ATCC is motion-triggered and works from video alone, which removes the need for inductive loops or piezo sensors and avoids pavement cutting during installation.

Do AVCC and ATCC work in fog, rain and at night? AVCC is largely unaffected because LiDAR and laser sensing do not rely on visible light. ATCC performance depends on camera specification, infrared or LED-assisted imaging, and how well the AI model has been trained on local conditions — a properly specified system handles all-weather, 24×7 operation.

Can both be integrated with an existing toll management system? Yes. Both should integrate with your Toll Management System, ETC/FASTag infrastructure, ANPR and command centre through open APIs. Confirm the integration method during evaluation rather than after award.