Work Location: Germany/Remote
Department: Remote sensing, Computer Science, Earth Sciences
Work load: 35 Hours / Week
Job Type: Master Thesis
General Description:
Precision agriculture is a cutting-edge approach to farming that utilizes advanced technologies and data-driven techniques to optimize productivity and efficiency while minimising environmental impact.
By integrating tools such as satellite imagery, drones, sensors, and GPS technology, various kind of data can be collected. After gathering the data and analyzing them using image processing, deep learning and GIS techniques, a wide range of Agricultural information/products can be provided to serve a broad spectrum of users like farmers, researchers and decision makers. That results in reducing costs, increase crop production enhanced crop yields, making precision agriculture a promising solution for sustainable and efficient food production in the face of increasing global challenges.
In Nabtaplaya we currently managed to build two solutions for agriculture which is crop classification and crop diseases.
Crop Classification:
In crop classification, we use a combination of optical and radar satellites like sentinel-2 and sentinel-1 to train our AI model in order to produce crop maps like for rice, corn, cotton, etc at 20 m resolution. The training data is created by the reference dataset provided by American Center for Spatial Information Science and Systems. The reference crop maps used in building our AI model is only available in USA. However, the models built can be also tested and validated over other areas in the globe if reference data is available. For example, a recently released crop type map model by ESA Cereal Model could be used for validation. Most of crop classification work was done in corn specifically in IOWA state in the USA (See Figure 2). An accuracy of 83% was reached. The model wasn’t including the temporal factor via using time series but this will be our kept for future work.
Crop Diseases:
While for crop diseases, Aerial imagery collected by drones were used. The training dataset contained information about crop diseases/issues like Weed cluster, Dry down, Nutrient deficiency and Endrow. The dataset is a public one and provided by Vision for Agriculture that host AI challenges. Each image consists of four 512x512 (10 cm) color channels, which are RGB and Near Infra-red (NIR). Each image also has a boundary map and a mask for the related problem. In order to reduce the computation power needed, the images were scaled to 256x256 (20 cm) resolution. Several AI models with different architecture were experiment on the data. Different results/performance due to different models and different crop issues were analyzed (See Figure 3).
Problem Solving:
In the frame of prototyping a Remote sensing solution, that is able to demonstrate a real life problem solving via combining the two solutions discussed above. To make it more specific, lets imagine a researcher/farmer, searching for corn fields in a specific region of interest and in a specific time of interest. After that, the researcher would like to check
which corn fields having lots of weed clusters. Weed clusters can affect corn fields and compete with corn plants for essential resources such as sunlight, water, and nutrients, which can significantly impact corn growth and yield. By this example, the problem solving is demonstrated in which we were able to locate areas where corn fields are affected. In addition, to measure the size of impacted areas. The same example could be applied to corn field and looking into factors like dry areas and nutrient defficenty, etc.
There is an initial UI design DEMO done on figma that explains the example above (See Figure 1).
Therefore, Its required to build a GUI interface using ArcGIS Experience Builder as described above and as shown in the DEMO video.
The GUI shall be able to acquire from user:
The area of interest on the map ( Square or Polygon)
The time of interest (Calendar)
The feature of interest:
Crop maps like corn/rice
Crop diseases like weed cluster, drought, nutrient deficiency
User selected polygon will be transformed into geo coordinates and saved with other inputs into a Jason file. This Jason file can be used to trigger some functions that we already implemented and running on our AWS platform. Once the Jason file is uploaded on AWS, the following will be executed:
Download Sentinel-2 and Sentinel-1 data based on user input
Co-register the downloaded images
Tile the images
Pass the data into our crop classification AI model that is exported in H5 format
Generate the crop map and save the results
After the Crop maps are generated, it should be downloaded from AWS and displayed on the map GUI. Afterwards the following shall be done:
Acquire the high-resolution imagery from ArcGIS maps for the user region of interest
Pass the acquired images to our crop diseases AI model that is exported in H5 format
Display the results of crop diseases on the classified crop map
By intersecting both crop solutions and building a GUI interface, a full END to END remote sensing proof of concept will be created. The END to END will cover lots of interfaces that demonstrates our general Remote sensing work flow architecture in real life problem solving.
N.b. The work will be organized on JIRA, in which the needed workflow will be broken down into tasks and subtasks. Jira will help in organizing the work, tracking progress, following up with different supervisors.
DEMO draft V1 - Figure 1
Corn Detection Over some Areas in IOWA STATE (2000 Km2) with 83% accuracy – Figure 2
Crop Disease results and performance using different models (Weed Cluster) – Figure 3
Essential Duties and Responsibilities:
Create a Graphical User Interface (GUI) using ArcGISExperienceBuilder with the following functionalities:
Gather Area of interest from ArcGIS maps
Gather Date of interest via calendar
Gather feature of interest
Export user request into a Jason file
Jason file will be used to generate crop maps
Download the crop maps and display results on the GUI Map
Acquire ArcGIS maps high resolution imagery to analyze crop map locations for crop diseases
Display the result on map
If possible to find on ArcGIS marketplace other agricultural products like irrigation, soil content, etc and add it to the GUI
Education and/or Work Experience Requirements:
Bachelor in computer science, remote sensing or related disciplines
Very good knowledge of GIS and Maps
Very good knowledge of ArcGIS and ArcGISExperienceBuilder
Good knowledge of Python
Basic knowledge of Image Processing
Basic knowledge of Cloud Technology
Very good verbal and written communication skills in English
Ability to work independently and to carry out assignments to completion within the parameters of instructions given, prescribed routines, and standard accepted practices
Timeline & Tasks:
Reading documentation for required tools : 1 month
Building an initial prototype GUI proof of concept for agriculture : 3 month
Research market tools that are open source and could be used to build GUI: 1.5 month
Thesis documentation and results: 2 month
Comments