HYDERABAD: On highways, slogans painted on trucks, often by hand, stretch across tailgates and registration numbers. For years, this visual chaos has defeated automated surveillance systems designed for neat, uniform plates.
It was this everyday disorder that pushed researchers at the International Institute of Information Technology Hyderabad to build Vahan Eye, a system designed not despite Indian roads, but because of them. The system has been tracking sand trucks and helping authorities curb illegal mining over some time now. The system’s success has now caught the attention of the Traffic Police. Officials are examining whether the technology can be extended to automatically detect traffic violations by two-wheelers, where missing, bent or non-standard number plates are a routine problem.
Vahan Eye was developed at IIIT-H’s iHub-Data after commercial Automatic Number Plate Recognition systems repeatedly failed to cope with Indian truck plates. While such systems work well in countries with standardised formats, Indian trucks tend to carry hand-painted numbers with uneven fonts, fading paint and creative spacing, often accompanied by decorative motifs or festival garlands.
The gap became apparent when the IT department approached the institute on behalf of the state Mineral Development Corporation, which needed a way to track sand-laden trucks and identify illegal movement. Imported systems were expensive and unreliable in field conditions.
Vahan Eye relies on open-source platforms
“Truck plates here may broadly follow black on yellow, but that’s where the similarity ends,” said Dr Veera Ganesh Yalla, CEO of iHub-Data and Adjunct Faculty at IIIT-H. “Every plate is different in style and execution. Off-the-shelf solutions are not built for that.”
Instead of forcing uniformity, the research team worked backwards from the problem. Drawing on earlier work at the Centre for Visual Information Technology, they rebuilt the recognition engine to handle handwritten characters and irregular layouts, and then packaged it into a deployable system that could run continuously on highways.
The first full-scale test came at Chityal on the Vijayawada–Hyderabad highway. Cameras were installed, cables laid and the system activated to track trucks entering Telangana. Each vehicle detected is checked against a whitelist of around 40,000 authorised sand carriers.
Since September, Vahan Eye has been running round the clock. It has had to contend with low light, dust, glare and plates partly obscured by decorations. These conditions routinely confuse conventional systems. The algorithm improves as it processes live traffic data, according to the developers.
Built by a team of fewer than five engineers, the system relies on open-source platforms and in-house models, avoiding the licensing and maintenance costs that can push commercial solutions into the tens of lakhs per camera.
It is this field-tested reliability that has prompted interest beyond sand transport. Traffic Police officials are now assessing whether the same approach can be adapted for two-wheelers, where enforcement cameras often struggle with damaged or unconventional plates.
If that happens, a system designed to follow sand trucks across highways may find a second life at city junctions.