**Machine Learning Based Tea Products Management System **
**Main Objective**
Implement the automation system for tea managing process which have done manually so far and analyze the future improvements according to algorithms
**Individual Research Questions**
1) We decided to create a system as a solution to the many problems of the declining tea industry in Sri Lanka. Here, the problem of the tea industry was that Sri Lanka was lagging behind in terms of technological development. The part that I will create in the system is supply management. The problem that affected the creation of such a component is that the tea pluckers are not able to correctly identify the age at which they should be plucked. This causes a lot of wastage of leaves. Also, it does not take a lot of time for people who are new to training. This way, after uploading a photo of the tea leaf, it can be determined whether the tea leaf is at the age to be pluck or not. Also, it is important for tea plantation owners to have an understanding of how the climatic conditions suitable for tea cultivation vary from area to area. For that, data is collected about two years ago and the algorithm is prepared.
2) Tea cultivation in Sri Lanka is widely spread and currently provides a lot of income to the country. But as of now, it has stalled. The reason for that is that there is no technical improvement in Sri Lanka. Due to that problem we decided to create a system. The component that I will create there is Measure the quality of tea powder. There we focus on the problem, after making tea powder, they are separated into quality as A, B, C. But industrialists have to spend a lot of time for that. Therefore, by uploading a photo as a solution to it, the system creates to indicate the quality of the tea powder. Then you can save a lot of time. Accordingly, he decided to create a solution based on that problem.
3) In the Tea Market, there’s no automatic way to get an idea to the seller or producer, about new trends and changing favors on customers.
4) When doing the the exporting process, It will be occur kind of wrong estimation, because of that it can be happen wastage.
Tea exporting brokers can requesting much tea vlume more than buyers expected
**Individual Objective **
1) Here we are waiting to get information from more than 100 tea plantation owners in a period of one year or more. Here the data of the tea estates belonging to more than 5 tea factories are collected. Here, the data we get from the plantation owners is the amount of leaves they picked (by weight), fertilizer applied to the tea plantations (in chemical organic form), sunlight time for the land, tea tree pruning time, rainfall during the year for every day or every day of this data year. I hope to get the data in a week. They are analyzed. It is useful for plantation owners for tea cultivation in the future. Also, by taking photos of the growing stage of tea leaves and determining the age of the tea leaves by image processing. In this section, we hope to get the appropriate age of the tea leaves for quality tea powder production. By using their mobile phones, plantation owners can take a photo of the tea leaves in their tea garden and upload it, giving them information such as the age of the tea leaves and the time to pick the leaves. This allows the planters to determine the date of pick-up by the harvesters. With the data obtained in this way, tea factories can produce more quality tea powder. In addition, they can get information about the amount of tea leaves they will receive for their production in the coming months. Waiting to get this information is in Badulla District and Nuwara Eliya District.
2) Every day, millions of people around the world enjoy the wonderful taste and health benefits of drinking tea. Therefore, the quality of tea powder is very important. Here they are divided into sections according to quality. The factors that measure the quality of tea powder are smell, touch, taste, nature of tea powder (like powder, granules). According to these factors, tea powder is divided into different segments.What we are doing here is analyzing the data about the quality of tea powder divided by image processing system. After that, by getting an image of the tea powder with the help of the user's mobile phone and uploading it to the system, it is possible to determine which category the type of tea powder in that image belongs to. This is more important for tea factory owners. It makes their work easier. This is also important for consumers who like tea a lot. we have made initial arrangements to use the tea board ,Tea Research Institute as well as several tea factories to obtain the quality classification data of this tea powder.
3) There are some major aspects to sell tea because customers have their own interests depending on their age, budget and health reasons. Packaging, weight, average price, flavors and herbals are some of them. So, in this component of our research, Local ‘Demand Management’ we are deeply looking into these aspects and help tea selling companies to expand their local market. To collect these data, we expect to visit few local famous supermarket chains around the country like Keells Super, Cargills Food City, Arpico Supercenter and do a questionnaire with customers and also with relevant supermarket employees with their permission. After collecting these data we’ll develop an algorithm from machine learning to help tea sellers to get an idea on what are the favorite aspects looking in a tea product when a customer wants to buy them, new trends and what need to be changed in production level to expand selling.
4) In Sri Lanka exporting tea products is processed by according to few factors. In next have to identify those factors which is related to tea exporting process. In here we consider tea categories, the way packeting them under this component. After data gathering according to tea product exporting, then analyze them using algorithm which target to get the future volume of tea product which tea manufacture companies/ factories are exporting. When gathering data collection, target is selecting time range and collect data related to that duration (2 years back). Data collection process will be taken from the selected tea manufacturing companies in Sri Lanka (Main resource of taking data collection comes under Sri Lanka Tea Board)
