ALPR Traffic Counting Service

The ALPR Traffic service combines traditional traffic counting with Automatic License Plate Recognition (ALPR). For each counted vehicle crossing your measurement cross-sections, the system captures and recognizes the license plate, providing detailed vehicle identification alongside traffic statistics.

Use Cases

• Origin-Destination (O-D) studies - Track vehicles between multiple measurement points
• Fleet monitoring - Track specific vehicle groups through measurement zones
• Border crossing analysis - Document vehicle movements between regions

Camera Requirements

The standard camera and mounting requirements described in Choosing a Camera and Mounting a Camera apply, together with the additional ALPR requirements below.

Daytime requirements: - Video frame rate: 25–60 FPS. - Optimal angle: 15–45° to the road surface.

Nighttime requirements: - A narrow-beam 850 nm IR illuminator, with an approximately 15–30° beam, mounted close to the camera axis with 10–30 cm separation. - Manual shutter speed: target 1/2000 s, minimum 1/1000 s, to eliminate motion blur under IR illumination. - A 2 MP varifocal camera, which performs better than 4K in low light because of larger pixels and less noise. - A black-and-white night profile with DNR, WDR, and BLC disabled because these filters reduce sharpness under IR. - Camera horizontal angle no greater than 20–30° and vertical angle no greater than 30°, so the plate is viewed close to head-on. - Illuminator power sufficient for a 15–20 m reading distance.

Note: Night ALPR is a measurement system consisting of the camera, IR illuminator, mounting, and configuration—not merely an application feature. Failed nighttime readings are almost always caused by hardware or optics rather than the algorithm. Contact us for a professional assessment and configuration guidance. The estimated gross cost of a complete night setup is approximately PLN 1,500.

Generated Reports and Data

The ALPR Traffic service produces comprehensive reports combining traffic counting with license plate recognition data.

Output Files

File	Description
History.csv	Complete vehicle history with all data fields
Alpr.csv	Filtered ALPR-specific data
Alpr0to500.xlsx, Alpr500to1000.xlsx, ...	Excel sheets with vehicle images and license plate crops (paginated for large datasets)
AlprOccurences.csv	License plate occurrence matrix grouped by time intervals
Quantity.csv	Standard traffic counting results
Report.docx	Generated measurement report

Data Fields

Each recognized vehicle record contains:

Field	Description
ALPR	Recognized license plate text
ALPR_score	Recognition confidence (0.0 - 1.0)
ALPR_type	License plate type classification
ALPR_region	Region/country code (e.g., "PL", "DE")
ALPR_lettercode	District/city letter code (e.g., "WA" for Warsaw)
ALPR_image	Cropped license plate image path
Image	Full vehicle image path

Recognition Statistics

The system calculates ALPR totals for quality assessment:

Metric	Description
Total vehicles	All counted vehicles in measurement
ALPR detected	Vehicles with license plates detected
ALPR read	Vehicles with successfully read plates
Region recognized	Plates with identified region codes
Letter code recognized	Plates with identified district codes

Polish License Plate Recognition Details

Polish standard license plate format: XX YYYYY or XX 123AB - First 2 characters: Region/district code (e.g., "WA" for Warsaw, "GD" for Gdańsk) - Remaining characters: Vehicle registration number

Common Polish region codes (ALPR_lettercode):

Code	Region	Code	Region	Code	Region
BA	Białystok	PZ	Poznań	TO	Tarnów
BN	Bielsko-Biała	RZ	Rzeszów	WA	Warsaw
DS	Dolnośląskie (Wrocław)	SL	Silesian (Katowice)	WR	Wrocław
GD	Gdańsk	SZ	Szczecin	WN	Włocławek
GO	Gorzów	SK	Świętokrzyskie	ZG	Zielona Góra
KA	Kalisz	PM	Pomeranian
KI	Kielce	KR	Kraków
KO	Koszalin	LD	Łódź
LE	Legnica	LU	Lublin

Plate type classifications (ALPR_type): - STANDARD - Regular vehicle plates (white background) - EU - EU format with country symbol - COMMERCIAL - Yellow plates (taxis, rental vehicles) - AGRICULTURAL - Green plates (farm vehicles) - DIPLOMATIC - Special format (embassies, international organizations) - TRANSIT - Temporary transit plates (red border)

Understanding ALPR_score in Polish context: - 0.95-1.0 = Excellent read (new, clean plates) - 0.85-0.95 = Good read (standard conditions) - 0.75-0.85 = Acceptable (typical Polish plates, some wear) - 0.60-0.75 = Marginal (dirty, faded, or obscured) - < 0.60 = Poor read (manual verification recommended)

Excel Reports with Images

For detailed analysis, the system generates Excel files containing:

• Vehicle thumbnails in the Image column
• Cropped license plate images in the ALPR_image column
• All metadata fields for filtering and analysis

Note: Excel files are paginated (default 500 rows per file) to ensure stability with large datasets.

Accuracy

Condition	Expected Accuracy
Daytime, clear view	85–95%
Rain or snow	80%
Dirty plates	50%
Night with IR and correct hardware configuration	50%
Night without IR or with unsuitable hardware	below 10%

Overall measurement error (including traffic counting): ~10%

---

Night ALPR

AITracker capabilities at night

• Transparent performance reporting — AITracker automatically generates 24-hour recognition charts showing actual performance for every hour. Failed reads are reported rather than hidden.
• Data for tender calculations — nighttime traffic statistics, typically 8–15% of daily traffic, help justify measurement tolerances in tender documents.
• Read-quality reporting — the ALPR report exposes recognition precision, plate types, regions, and letter codes; verify plate size and camera settings during installation.
• Camera configuration — configure exposure, night mode, and IR filters on the camera itself; AITracker does not control these settings.

Technical limitations

• Missing input cannot be reconstructed — if the image is blurred, overexposed, or lacks plate retroreflection, there is no information for the algorithm to read.
• Night ALPR is usually not an algorithm problem — unsuccessful reads are generally caused by unsuitable hardware, such as a wide-angle IR illuminator, a small-pixel camera, enabled DNR/WDR filters, or incorrect mounting.
• Software cannot compensate for unsuitable hardware — a weak camera, inappropriate IR illuminator, or incorrect angle must be corrected in the physical installation.

“The problem is in the optics and hardware, not in the algorithm.” — Sławomir Paszko, AISP

Hardware requirements for night ALPR

Camera

Parameter	Requirement
Shutter	Manual; target 1/2000 s, minimum 1/1000 s
Resolution	2 MP varifocal, preferred to 4K in low light
Night mode	Black and white; DNR, WDR, and BLC disabled

IR illumination

Parameter	Requirement
Wavelength	850 nm
Beam angle	Narrow, approximately 15–30°
Distance from camera axis	10–30 cm
Effective range	At least 15–20 m

Camera mounting

Parameter	Requirement
Horizontal angle	No greater than 20–30° to the path of travel
Vertical angle	No greater than 30°
Position	Close to head-on to the path of travel, not at a large lateral angle