en.Wedoany.com Reported - SunTec India, an Indian data services company, has completed a data annotation project for a government infrastructure consulting firm, processing over 3 million annotations with 99% accuracy. The dataset will be used to train a pavement asset and damage detection model covering India's national and state highway networks.
The annotation team used bounding boxes, as well as 4-point and multi-point polygon annotations on the client-hosted CVAT platform to label highway images, identifying pavement damage and classifying road assets according to IRC82 government standards. The project covered over 1,000 kilometers of highway corridors, involving 71 different damage and asset categories. Among these, 4-point annotations accounted for approximately 80% of the total workload, while the remaining 20% required highly detailed multi-point annotations.
The image annotation project initially began as a small-scale pilot with 4 annotators and 1 reviewer. After the pilot was completed, the scope rapidly expanded, and the team grew to 35 annotators and 7 reviewers, all domain experts with backgrounds in civil engineering. Accuracy remained stable despite an eightfold increase in team size. SunTec India attributes this achievement to a reasonable ratio of reviewers to annotators, as well as systematic training on the IRC82 standards before annotating real data.
Rohit Bhateja, Director of Digital Engineering Services and Head of Marketing at SunTec India, stated that what he is most proud of is that accuracy remained unchanged as the scale expanded. He added that highways are not uniform, with varying weather, lighting, and wear conditions across different regions. Annotating 71 categories over a thousand-kilometer range generates thousands of edge cases, and achieving 99% accuracy under these circumstances demonstrates that human-machine collaborative workflows remain stable regardless of dataset size.
This milestone solidifies SunTec India's position in professional data annotation for artificial intelligence and machine learning applications, particularly in geospatial and aerial image annotation. It also reflects the growing market demand for specialized AI training data with technical annotation guidelines and dense categories. The company expects further increases in demand for such high-consistency annotations in regulated industries.
Founded in 1999, SunTec India is an AI-based IT and digital services company that provides intelligent data services and AI technology platforms for AI training, ESG research, B2B sales intelligence, and document processing, as well as data management operations such as data collection, cleaning, and enrichment. With over 25 years of industry experience, the company has served more than 8,500 clients across over 50 countries and employs over 1,500 full-time professionals.
