AN EFFICIENT CROP HEALTH MONITORING SYSTEM USING AERIEL IMAGES
Agriculture is considered the backbone of Pakistan's economy, which relies heavily on its major crops. This sector directly supports the country's population and accounts for 24 percent of gross domestic product (GDP). [1]
2025-06-28 16:30:13 - Adil Khan
AN EFFICIENT CROP HEALTH MONITORING SYSTEM USING AERIEL IMAGES
Project Area of Specialization Artificial IntelligenceProject SummaryAgriculture is considered the backbone of Pakistan's economy, which relies heavily on its major crops. This sector directly supports the country's population and accounts for 24 percent of gross domestic product (GDP). [1]70% population of this country lives in rural area and almost 45% of them are somehow engaged with agriculture. [1] But there are vast gaps between the acquired and actual output of produce, which suffers due to a lack of appropriate technology, use of inputs at improper times, unavailability of water and land use and mainly because of various diseases. This usually occurs due to late identification of the pests, since the cultivated region is spread over acres of land. It not only negatively affects the produce but also significantly reduces the amount of produce. So, there is a dire need to propose some effectual well-being strategies to decrease the effects of pests. But traditional methods are extremely time-consuming and require a human resource for visually observing the plant leaf patterns and diagnose the anomaly. Farmers have to manually survey whole area for defects in the crops and then have to spray pesticides all over the crops which have very damaging effects for the crop yield. This is time consuming and as well as very difficult. The time consumption as well as difficult process of examining the fields can be made easy with use of drone with digital camera for survey of the area. This drone can be program to survey the selected area and capture images along its path. After the drone has completed its survey, images or video captured by the drone can be used to defect detection using semantic segmentation, which helps the farmer to spray pesticides on a specific area.
In this proposed system we’ll be using the aerial images; aerial images will be obtained with the help of drone. Image Pre-processing and Identification of the affected areas will be our main focus in this project. After some initial pre-processing, images then stitched together by means of image stitching.Firstly, we’ll identify the region of interest through scanning which will based on the crop appearance and crop growth, the most obvious indicator of crop health is their general appearance. Lightness or discoloration in foliage or abnormal texture or spots on the leaves. Then we’ll implement the basic framework in MATLAB and then selection of filters will occur. After this actual implementation of framework will start to train the system. For testing purpose real time aerial images or unmarked dataset will be used.
References
Project ObjectivesTo implement a secure and reliable system that aims to monitor the health of the crop, identification is based on the color, growth, texture and size. This can be obtained by passing through different stages listed as follows:
- Using dataset of aerial images
- Stitching of the images on aerial images and scanning.
- Determining region of interest.
- Comparing attributes based on that will declare the affected areas in a broader sense.
Most of the Pakistan’s economy rely on agriculture as it contributes 24 percent to GDP, so identifying any unhealthy crop in early stages is very crucial as these plants causes a large drop in the production and economy of the farmers. Therefore, our system will consist of following features:
- The Aerial images are acquired from the environment using a Drone. Note that we will not be addressing the issues of the drone. Dataset of aerial images will be provided by our internal and external. Our main focus will be on the Image Pre-processing and Identification of the affected area. We will be addressing the mortalities of aerial imaging.
- Images are collected from dataset are stitched together to obtain a single high resolution image. Image stitching is the process of combining multiple photographic images with overlapping fields of view to produce a segmented panorama or high-resolution image. The image is then scanned to identify region of interest. We will do crop profiling, with consultation of expert from agriculture Dr. Mashooque Ali Talpur who is an Associate Professor and chairman faculty of Agriculture Engineering at Agriculture University Tandojam Sindh.
- Affected areas are Identified based on following attributes:
- Crop appearance
The most obvious indicator of crop health is their general appearance. Lightness or discoloration in foliage or abnormal texture or spots on the leaves and the size
- Based on these attributes we will declare the area healthy or unhealthy.
Figure 1 Flow chart of Monitoring Crop Health system
Data Collection
We’re collaborating with Rafiq Hakim who has 100 acres farm. Data Set will be collected from his farm and will be provided by our Internal and External. Location of the farm is Village Abdul Hameed Alwani, Deh Karampur, Union Council Karampur, Taluka Mirpur Sakro, District Thatta. Hence we’ll be targeting local crops. Data Set that will be available to us will be local and we will target seasonal crop.
Equipment/Tools
Hardware: Workstation. Software: Matlab, Mongo DB, React, PHP, Python.
Benefits of the ProjectFarmers must always aware of the current health of crops and must be prepared to address any problem with solutions that don’t end up causing more and by implementing our project farmers get the current update of crop’s health
The major benefit of our project is the continuous checking of the crop’s growth. Farmers can not able to check each crop the whole day which is done by our drone which takes a new image at every rotation.
Moreover, some farmers become overly reliant on insecticides and other chemicals to eliminate their pest problems a grievous error, as this will likely lead to even more serious problems. Even the indiscriminate application of mineral fertilizers may inadvertently boost pest populations by making conditions ideal for them to thrive.
Technical Details of Final DeliverableAerial imaging provides a landscape view of crop fields that can be utilized to monitor crop health. In this proposed system we will be using DJI Mavic Mini Combo - Drone FlyCam Quadcopter to captures the crop of 100 acres field located in Village Abdul Hameed Alwani, Deh Karampur, Union Council Karampur, Taluka Mirpur Sakro, District Thatta. Dataset collection is the mileston planned in project.
Image mosaicking, it is basically blending together of several arbitrarily shaped images to form one large radiometrically balanced image so that the boundaries between the original images are not seen and is performed by detecting keypoints and extracting local invariant descriptors (SIFT, SURF, etc.) from two input images. Matching the descriptors between the images. Using the RANSAC algorithm to estimate a homography matrix using our matched feature vectors.
UNET architecture is used to perform semantic segmentation to label the images and form segments based crop appearance which will be further used to classify the crop as healthy or unhealthy. Nvidia Jetson neno 2gb development kit will be required for faster processing.
The implmentation of this system will help the farmers to monitor the large crop fiels efficiently which will increase both the quality and quantity and in this way we can utilize human experties in a best manner.
Final Deliverable of the Project Software SystemCore Industry AgricultureOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Decent Work and Economic Growth, Responsible Consumption and ProductionRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 76000 | |||
| DJI Mavic Mini Combo - Drone FlyCam Quadcopter UAV | Equipment | 1 | 60500 | 60500 |
| Nvidia Jetson neno 2gb development kit | Equipment | 1 | 7500 | 7500 |
| Printing and Stationary | Miscellaneous | 1 | 5000 | 5000 |
| Farm Visit | Miscellaneous | 1 | 3000 | 3000 |