SunTec India Achieves Major Milestone in Highway Infrastructure Data Annotation Project
In a remarkable feat, SunTec India has successfully concluded a significant data annotation initiative aimed at enhancing highway infrastructure monitoring in India. The project, which is backed by the government, saw the company process over 3 million image annotations, achieving an impressive accuracy rate of 99%. This endeavor is intended to facilitate the training of advanced models capable of detecting damage in highway assets across the country's expansive national and state highway networks.
The project operated on a client-hosted Computer Vision Annotation Tool (CVAT) platform. The SunTec team meticulously classified and labeled highway images using techniques such as bounding boxes and four-point/multi-point polygon annotations. These methods were essential for identifying road surface damage and classifying various road assets following the Indian Road Congress (IRC82) standards. In total, 71 distinct categories were developed to cover an extensive 1,000-kilometer corridor of highway network.
Analyzing the workload, approximately 80% consisted of four-point annotations, while the remaining 20% required the application of exceptionally detailed multi-point annotations, which added layers of complexity to the task. This image annotation project began as a pilot involving just four annotators and one reviewer, as a proof of concept. However, following its initial success, the team’s size expanded drastically to accommodate 35 annotators and seven reviewers, all of whom were seasoned subject-matter experts with backgrounds in Civil Engineering.
Notably, throughout this substantial increase in team size, the company maintained its remarkable accuracy rate. Rohit Bhateja, the Director of Digital Engineering Services and Head of Marketing at SunTec India, expressed his pride in the team's performance, stating, "What I'm proudest of is that the accuracy held as the scale grew." He further elaborated, emphasizing the challenges of labeling data for a highway context where variations in weather, lighting, and wear affect the uniformity of assets across regions. The complexity of labeling 71 categories over 1,000 kilometers produced thousands of edge cases, highlighting the robustness of their human-in-the-loop workflow.
This accomplishment not only solidifies SunTec India's status as a leader in domain-specific data labeling for artificial intelligence and machine learning applications but also underscores the increasing demand for specialized AI training data. The company's success reflects a broader trend towards high-consistency labeling across regulated fields, with a growing expectation for technical annotation guidelines and dense categorization.
Founded in 1999, SunTec India boasts a wealth of experience in the IT and digital services sector, particularly in providing intelligent data services for various AI and tech platforms, aiding in AI training, ESG research, B2B sales intelligence, and document processing. With over 25 years of expertise, the company has served more than 8,500 clients across over 50 countries and employs a talented workforce of over 1,500 professionals. Their reputation for excellence continues to attract international clients in search of superior data management solutions.
As the demand for specialized annotation increases, SunTec India stands poised to leverage its extensive capabilities to meet future requirements, affirming its role at the forefront of the evolving landscape of AI training data services.
The project operated on a client-hosted Computer Vision Annotation Tool (CVAT) platform. The SunTec team meticulously classified and labeled highway images using techniques such as bounding boxes and four-point/multi-point polygon annotations. These methods were essential for identifying road surface damage and classifying various road assets following the Indian Road Congress (IRC82) standards. In total, 71 distinct categories were developed to cover an extensive 1,000-kilometer corridor of highway network.
Analyzing the workload, approximately 80% consisted of four-point annotations, while the remaining 20% required the application of exceptionally detailed multi-point annotations, which added layers of complexity to the task. This image annotation project began as a pilot involving just four annotators and one reviewer, as a proof of concept. However, following its initial success, the team’s size expanded drastically to accommodate 35 annotators and seven reviewers, all of whom were seasoned subject-matter experts with backgrounds in Civil Engineering.
Notably, throughout this substantial increase in team size, the company maintained its remarkable accuracy rate. Rohit Bhateja, the Director of Digital Engineering Services and Head of Marketing at SunTec India, expressed his pride in the team's performance, stating, "What I'm proudest of is that the accuracy held as the scale grew." He further elaborated, emphasizing the challenges of labeling data for a highway context where variations in weather, lighting, and wear affect the uniformity of assets across regions. The complexity of labeling 71 categories over 1,000 kilometers produced thousands of edge cases, highlighting the robustness of their human-in-the-loop workflow.
This accomplishment not only solidifies SunTec India's status as a leader in domain-specific data labeling for artificial intelligence and machine learning applications but also underscores the increasing demand for specialized AI training data. The company's success reflects a broader trend towards high-consistency labeling across regulated fields, with a growing expectation for technical annotation guidelines and dense categorization.
Founded in 1999, SunTec India boasts a wealth of experience in the IT and digital services sector, particularly in providing intelligent data services for various AI and tech platforms, aiding in AI training, ESG research, B2B sales intelligence, and document processing. With over 25 years of expertise, the company has served more than 8,500 clients across over 50 countries and employs a talented workforce of over 1,500 professionals. Their reputation for excellence continues to attract international clients in search of superior data management solutions.
As the demand for specialized annotation increases, SunTec India stands poised to leverage its extensive capabilities to meet future requirements, affirming its role at the forefront of the evolving landscape of AI training data services.
Topics Consumer Technology）
【About Using Articles】
You can freely use the title and article content by linking to the page where the article is posted.
※ Images cannot be used.
【About Links】
Links are free to use.