Nasscom Community

Data Annotation Landscape and Trends: The India Story

2 Mins read

We are well aware of the Achilles’ Heel of AI, training data on which the model is built, it has the power to make or break the AI model and the subsequent decisions it takes. My previous article highlighted the importance of data annotation services in creating well-labelled training data for enterprise AI models. The article also talks about the building blocks of successful service delivery in the annotation space and the business models prevalent globally.

Image Source: Analytics India Magazine

Let us in this article try to get a deeper understanding of the India story with respect to the data annotation landscape and a few trends that are driving the India market. The data annotation market serviced by India in FY20 is ~USD 250 Mn and the cost billed to the client for any project is based on the combination of the kinds of use cases, number of data items to be labelled, complexity and type of annotation required. The infographic below highlights the India market dynamics.

Source: Data Annotation – Billion Dollar Potential Driving the AI Revolution

The same business models as mentioned in my previous article apply in the Indian context as well – crowdsourced platforms, managed service providers (MSPs) and SaaS offerings. However, the managed services model drives the majority revenues approximately 65-70% of the total revenue.

Source: Data Annotation – Billion Dollar Potential Driving the AI Revolution

The Indian MSPs follow different workforce models with majority of the employees represented by non-metros:

  1. Dedicated Workforce Model
  • Service providers with dedicated full-time employees that utilize the internal annotation tool
  • The dedicated workforce is the USP and is trained to deliver client projects at the highest level of quality
  1. Partnered Workforce Model
  • This model is adopted by service providers that follow a platform-first approach with the tool serving as the USP
  • The workforce comes from verified BPM partners from Tier II/Tier III cities who are trained on the annotation platform

MSPs also impart an average of 4 weeks of training to the hired workforce and look for learnability instead of education.

Training – Apart from the training provided prior to live projects, employees receive an average of 2-4 weeks of training through the year

Mindset over Education – While basic intermediate education is sufficient, the key attribute while hiring is mindset and ability to learn

Employee Earning – The monthly salaries offered by MSPs can go up to USD 300 per month for generic data annotation tasks, for sector-specific expertise it could be even higher.

Source: Data Annotation – Billion Dollar Potential Driving the AI Revolution

Watch out for my next article that provides more details about the data annotation industry. Download our full report on Data Annotation – Billion Dollar Potential Driving the AI Revolution for more interesting insights.

The post Data Annotation Landscape and Trends: The India Story appeared first on NASSCOM Community |The Official Community of Indian IT Industry.