by admin
The Computer Science Program Universitas Pendidikan Ganesha has entered into a collaboration with KGEO for research collaboration. Two research projects have been proposed as part of this collaboration. These research projects are submitted in the form of international collaborative research. The titles of the research projects are as follows:
Title : Auto-Mapping and Predicting Crop Production
Project investigator: Kadek Yota Ernanda Aryanto, S.Kom., M.T., Ph.D.
Team Members:
Dr. Komang Setemen, S.Si., M.T.
Dr. Luh Joni Erawati Dewi, S.T., M.Pd.
This research is planned to be conducted over multiple years. In the first year, the research will focus on satellite image processing for Land Use/Land Cover (LULC) classification, integrating meteorological data and remote sensing. The key objectives include identifying significant trends during three periods: the early season (SOS), peak season (POS), and late season (EOS). The second year will emphasize yield prediction and mapping using data obtained in the first year, with spatial mapping for each prediction. Meanwhile, the third year will concentrate on fully automated crop production mapping and prediction, delivered through a web-based Geographical Information System (Web-GIS).
Title : Object-Based Image Classification of Economic Crops with Machine Learning Methods
Project Investigator: Dr. Luh Joni Erawati Dewi, S.T., M.Pd.
Team Members:
Kadek Yota Ernanda Aryanto, S.Kom., M.T., Ph.D.
Dr. Gede Indrawan, S.T., M.T.
This research aims to develop a mobile application capable of automatically annotating images captured from mobile device cameras based on crop data, geolocation, and other necessary information. Machine learning will be employed to accurately classify plant types. There are challenges in this work as the entire process must be rapid yet lightweight to be adequately performed on mobile devices. This project will focus on four economic crops in Thailand: rice, sugarcane, cassava, and maize.
The collaborative research provides opportunities for faculty and students to intern at the research site. Several students and faculty members are participating in internships and sub-research projects. The faculty and students involved in internships are as follows:
Name: Ketut Agus Seputra, S.St., M.T.
Project Title: Developing a Backend App to Provide Timeseries Vegetation Condition Index With Python Geemap
The project’s objective is to create an app that supports surveyors in conducting crop distribution surveys. The machine learning feature will automatically identify the crop type and provide geolocation tagging. This feature will be complemented by an interactive map for pinpointing locations. Geolocation tagging will provide coordinated position information used in remote sensing. Remote sensing will analyze field information and Vegetation Condition Index (VCI) data, including Elevation, Average Temperature, and Water Resources, depending on geolocation. Accurate image capture coordinates are crucial for remote sensing analysis. To ensure a seamless experience, the app will include an interactive map display with zoom and pan features. The information will be presented in an interactive Geospatial Information System (GIS) interface.
Name: Kadek Prima Giant Marta Dinata Project
Title: [Title Not Provided]
The project aims to compare three different machine learning algorithms in the context of classifying Land Use Land Cover (LULC). LULC is a crucial aspect of mapping and understanding land usage and cover, including categories such as forests, agriculture, urban areas, etc. The three algorithms to be compared in this project are Support Vector Machine (SVM), Random Forest (RF), and a manual approach using Maximum Likelihood. SVM is a powerful algorithm for separating different data classes by finding the best hyperplane. RF is an ensemble algorithm capable of handling data complexity and performing well in classification. The manual approach with Maximum Likelihood serves as the traditional baseline method, focusing on calculating class probabilities based on the statistical distribution of training data.
Name: Komang Harry Sudana
Project Title: Development Of Machine Learning Models For Prediction of NDVI Values On Time Series Data In The Framework Of Crop Phenology Analysis in Paddy Crops.
This research aims to develop a prediction model for NDVI (Normalized Difference Vegetation Index) values using time series data, with a primary focus on analyzing cropping patterns in rice plants. The dataset for this research will be obtained from Sentinel-2 and MODIS (Moderate-Resolution Imaging Spectroradiometer) satellite images, covering a three-year period from 2020 to 2022. The dataset will be sourced from the ESA (European Space Agency) platform and Google Earth Engine, which facilitate efficient data retrieval. Advanced time series analysis methods and modeling techniques will be applied to predict NDVI values with high accuracy. The study results are expected to provide a comprehensive understanding of vegetation changes in rice farms over the studied timeframe. This information will be instrumental in supporting decision-making in farm management, crop growth monitoring, and environmental monitoring on a broader scale.
Comments are closed.
