Photo montage: Ajay Mohanty
India’s tea industry is turning over a new leaf — and your morning cuppa may just get more aromatic and flavourful as a result. Inspired by the success of other sectors such as FMCG, manufacturing, banking and telecom in using the latest technologies to improve efficiency, the Tea Board of India and other tea bodies are looking at artificial intelligence (AI), big data and analytics and machine learning (ML) to enhance the quality of tea so that it can command higher prices in the world market.
A labour intensive sector where wages account for over 50 per cent of the total production cost, the tea industry has traditionally been resistant to technological change. However, it has been under severe financial stress in recent years. While some large tea producing firms blame government policies for this, others complain that the sector has been hit hard by the shrinking size of the industry and Indian teas fetching lower prices in private sales as well as auctions.
Insiders, including the Tea Board of India and some exporters and auctioneers, feel that to combat the problems that the industry faces today, the first thing that needs to be done is to improve the quality of Indian tea. This will not only enable producers to get better prices for their teas, but will also boost the reputation of Indian tea in the global market. Indian tea has taken a beating ever since the market has been flooded with quality teas at competitive price points coming from countries like Sri Lanka and Kenya.
Now the industry is planning to use technology
to try and address the issue of quality. The Tea Research Association (TRA), the technological backbone of the industry, has teamed up with startup firm Agnext Technologies and IIT-Kharagpur to develop an AI, ML and data analytics-enabled system to count the fine leaf percentage of plucked green leaves before they are sent for processing. Fine leaf, which refers to two young leaves and a bud, is the single-most crucial factor that determines the quality of tea. The higher the percentage of fine leaves, the better the tea quality and higher the price.
The TRA mandates that teas should have at least a 60% fine leaf count
The system, which consists of a box-like equipment as well as a mobile app, is currently at a pilot stage. The equipment is armed with various types of sensors and cameras that are connected to the internet via a software platform. Once they are commissioned, the units will be installed in tea processing factories. The plucked leaves will be placed in them for the fine leaf to be counted.
As for the mobile app, users need to install it on their phone, click a picture of the tea leaves on the app, and send it to the TRA.
In both processes, the data sent by the equipment and the app go to the back-end where it is tallied with the data collated by IIT Kharagpur and TRA using computer vision technology
and AI. The result not only shows up the fine leaf count of the teas, but also the plant variant, its health and various other indicators.
“Hence, the user knows if the harvest has enough fine leaves to produce quality tea,” says Joydeep Phukan, secretary, TRA. The mobile app is to have two interfaces — one for the producer and the other for the tea factories, he adds. “So both the producer and the factory owner will know the quality of the plucked leaves before these are processed into tea,” he says.
Though the TRA mandates that teas should have at least a 60 per cent fine leaf count, tea estates often fail to maintain this standard. Hence, once the app and the equipment are in place, they will also help the Tea Board inspectors keep tabs on what individual gardens are doing to maintain quality and whether or not they are abiding by the TRA norms.
Moreover, since the counting is currently done by hand, the process is open to manipulation and tampering. “Since it is automatic, the new technology
will remove any anomalies in the fine leaf count,” says Phukan.
The app has now achieved a 55 per cent accuracy rate. The TRA wants to launch it once it achieves a consistent success rate of 85 per cent.
The Tea Board too is doing its bit to give the industry a technological fillip. It is coming out with an app aimed at the small tea growers. The Board will use the app to issue advisories, disseminate information related to its various schemes and provide real-time weather updates.
Experts are upbeat about the industry adopting smart technologies. ”It's time the tea industry started using modern technology to address the industry's key concerns,” says A K Roy, deputy chairman at the Tea Board.
Atul Asthana, who heads the engineering committee at TRA and is also managing director at the Goodricke Group, says the smart equipment can be used by individual gardens for their internal quality checks and will help bring transparency into the trade. “This way, we will know the status in individual gardens, what aspect of the trade needs improvement, and so on.”
Insiders say that the move to adopt cutting edge technologies could help the tea industry tackle the problem of low prices, thereby offsetting the sluggishness that has set into the business. “The prices of tea have not been rising to the extent needed to meet the cost of production. While prices are left at the mercy of the market, the overhead costs are fixed,” says Azam Monem, director, McLeod Russel, the world’s largest tea producer.
But if the quality of Indian tea soars, prices too will go north. It remains to be seen if that will happen, and if digitisation in tea — the first such attempt in the world — does indeed succeed in improving the taste of the brew that goes into your cup.
Tech in a tea cup
AI, ML and data analytics-enabled system is used to count the fine leaf percentage of plucked leaves
The higher the percentage of fine leaves, the better the tea quality
The counting is done via a box-like equipment armed with sensors and cameras
Plucked leaves are placed in the box for the fine leaf to be counted
Planters can also click photos of tea leaves on a mobile app and send them to the Tea Research Association for quality analysis