Automated Yield Estimation of Kinnow
Kinnow is considered to be one of the best commercial fruits of Pakistan. It is cultivated on a large scale in Punjab Province and make a huge contribution to the overall economy of Pakistan. Pakistan generate pretty good revenue from the exports of Kinnow and the production of Kinnow is 60% in the
2025-06-28 16:30:24 - Adil Khan
Automated Yield Estimation of Kinnow
Project Area of Specialization RoboticsProject SummaryKinnow is considered to be one of the best commercial fruits of Pakistan. It is cultivated on a large scale in Punjab Province and make a huge contribution to the overall economy of Pakistan. Pakistan generate pretty good revenue from the exports of Kinnow and the production of Kinnow is 60% in the total of citrus fruit production. Pakistani firms estimate to produce 2.2 million tons of kinnow during the current season and will export 20 percent of it and earn a good ratio of profit from it. To generate maximum profit from Kinnow export we need to have an estimate of kinnow production.
At the present, the estimate is performed manually which include issues of inaccuracy, inefficiency and cut down the profit. Labor is required to harvest Kinnow at a certain time. Not having an estimate of yield will lead to having less labor that will consume a lot of time and affect the market dealings. In Pakistan most of the dealers agree to make a deal according to the trees having visibly less quantity of kinnow that results into a huge loss to the farmer as there might be other trees that have a large quantity of fruits on it than the tree that was standardized.
Our aim is to develop an automated system to estimate the yield of kinnow to help farmers to hire required labor for harvest and prevent time and cost constraints. Farmers having an estimate of kinnow can help them to make dealings in the market beforehand to earn profit. An approach for yield estimation is to collect the dataset of images with the help of drone. Images are preprocessed to enhance information. Image processing is applied to identify objects (kinnow) from given dataset. Count of objects (kinnow) is estimated through Machine Learning Algorithms.
Project ObjectivesOur objective is to provide the farmer with an automated yield estimation system.
Cost: Our system will give estimation of yield of the kinnow and help farmer to make dealings with the vendors beforehand and earn huge amount of profit.
Accuracy and Efficiency: Our system will be automated to improve overall efficiency and accuracy of estimating the yield, and help the farmers to gain maximum profit from their orchards.
Image Processing/ Machine Learning: Our system will have high accuracy and efficiency rate by using Image Processing and Machine Learning algorithms
Project Implementation Method- To get the dataset of tree images we will have to travel to Sargodha to visit Kinnow orchards and after taking permission from orchard’s owner we will able to gather dataset in form of the images. We might have to go through different trials to collect suitable images for our dataset. We might also have to collaborate with Citrus Research Institute, Sargodha for data collection phase. Our dataset will be collected through drone at an angle of 450.
- Algorithms of Image Processing will be applied on dataset to enhance information in images. Noise reduction will be done before further processing. Image Stitching will be applied on dataset to remove repetition of objects in multiple images.
- Machine learning algorithms will be applied to detect required objects form dataset. We will learn different models of deep learning and will use the suitable one for estimating the count of Kinnow from collected dataset.
- Regression techniques will be used to estimate count of kinnow on inner side of tree that were not visible to camera and was not recorded in dataset.
- The outcome of our system will be the estimated count of kinnow available on a tree.
Estimation: By this system, the farmer will have an estimate amount of his fruits. And that can help in future dealings with his customers.
Cost Advantages: Having an estimate of the yield can help the farmer to manage market dealings with vendors and to earn a handsome profit by managing resources beforehand.
Low Cost Solution: Our system provides a low cost solution for estimating yield for the farmers. By proper use of this system and knowing yield will help the farmer to get maximum profit.
Dying Plants: it is easy to identify dying plants through our system. Average age of Kinnow tree is 50 years, and if yield of tree is really low this means that the particular tree is dying and it would be better time to plant a new tree. So it would help to know when to remove old plants and sow new ones.
Efficiency and Accuracy: Through its system properties, it can provide a higher efficiency and accuracy as the manual labor can miscalculate and can be at more error.
Technical Details of Final DeliverableA software that will give estimate count of the kinnow on the tree using different methods of image processing and machine learning. It will estimate yield by gathering dataset of the images, and then segmenting the images to identify kinnow. Techniques of Machine Learning (Regression) will be used to tell approx. count of kinnow which is not visible in images (kinnow in inner side of tree)
Final Deliverable of the Project Hardware SystemCore Industry OthersOther IndustriesCore Technology RoboticsOther TechnologiesSustainable Development Goals Decent Work and Economic Growth, Partnerships to achieve the GoalRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 79900 | |||
| Clamp Meter UT207A | Equipment | 1 | 9000 | 9000 |
| SD Card 32GB (Kingston SDCA10) | Equipment | 2 | 2200 | 4400 |
| Phantom 4 Battery | Equipment | 1 | 28000 | 28000 |
| Gimbal (HAKRC) | Equipment | 1 | 8500 | 8500 |
| Phantom 4 Quick-Release Propellers (1CCW+1CW) | Equipment | 4 | 2500 | 10000 |
| Variable Power Supply YH3010D | Equipment | 1 | 10000 | 10000 |
| Trip to Sargodha | Miscellaneous | 2 | 5000 | 10000 |