I am awarded the Erasmus Mundus Joint Master Degree (EMJMD) Programme for Copernicus Masters in Digital Earth (CDE) for 2020. I am studying Earth Observation and Geoinformatics at the University of Salzburg in my first year of the master's program, followed by GeoData Science at Universite Bretagne Sud in my second year.
The purpose of this page is to track a few of my tasks, I accomplished during my master studies in Austria and France.
Orientation session was conducted on the 1st week of our master program. It was dedicated to introduce ourself, the structure of curriculum and student life at CDE partner universities. Lectures were provided on Career development; planning and design of each module; structuring of documents according to media and target audience; introduction to licenced software and tools and; elementary research design and professional ethics.
It was a fun week, as we "second batch of CDE" shared our personal and professional experiences. We were 15 people from 13 different countries around the world. Despite the fact that some of us were unable to physically be present in the room due to the pandemic, we were all virtually connected. However, 11 of us were able to enjoy Salzburg before its second lockdown and attend few weeks of classes in physical mode. This marked the beginning of a new chapter in our lives.
I submitted this ppt as for the Orientation Project.This course covered the full workflow from image acquisition (new sensor types and devices), over advanced pre-processing and pre-classification techniques, and object-based image understanding including quality assessment.
At the end of the course, I acquired knowledge on earth observation and its application, especially in the context of the Copernicus programme. I was familiarized with advance methods, analysis tools and techniques for tackling real-world scenarios with remotely sensed imagery. For the first time, I learned more about OBIA (Object-Based Image Analysis), as a part of this course. It broadened my vision of Knowledge-Based Classification and Class Modeling. The following are just a handful of the numerous topics covered in this course:
This was a seminar course in which participants were given the option of selecting and performing analysis and modelling on a specific topic using remotely sensed data. The seminar was divided into 4 different phase: Image acquisition, pre-processing, image analysis and cloud processing.
At the end of the course, I was more familiarized with spatial analysis tools and techniques. I receive an improved understanding of the availability, suitability, and applicability of tools and analytical methods in the field of Earth observation with a focus on remote sensing and image analysis techniques. Though my chosen topic was machine learning, I equally learned Image tasking of very high-resolution imagery; in-depth understanding on radiometric and top of atmospheric correction; clear differentiation on DSM and DEM; multi-resolution segmentation and knowledge-based classification on Sen2cube datacube; CNN with geospatial flavour; Dynamic time wrapping and processing in GEE.
In order to complete the assigned task, I gave a presentation and submitted a seminar paper.
The course work for this topic was divided in two different module: one covering the theoritical part and the other was more focused on the practical sessions.
The session imparted a basic understanding of software engineering fundamental principles, notably their connections to proven geographic methodologies and Geographic Information Systems (GIS). A particular focus is dedicated to object-oriented programming with Java and web-based technologies for geographic applications. From this course I learned simple aspects of software engineering in the context of geoinformatics domain. With this course I used JAVA programming language for the first time and maybe since I was familiar with few other programming, it was handy for me. Though handy but still new and very useful for me for working with spatial data and databases. As part of the practical session, I teamed up with Eike Blomeier and Ella Christie to design a JAVA program for Visualizing Tweets in Google Earth Pro.
The course provided the knowledge that enabled us to read and implement geospatial data models based on standardized techniques. Here, Geospatial Data Models were disscussed as the foundation of efficient geospatial data processing since it impacts service interfaces, performance and the ability to flexibly exchange data. Starting from a quick review of modeling basics, it included discussions and hands on experiences in UML, GML and the General Feature Model.
With a focus on spatial awareness and literacy, for effective operations in humanitarian action geospatial technologies are considered as a crucial ingredients. This course was introduced as a seminar session where Geospatial tools, including web and field mapping, Earth browsers and VGIs, VHR satellite data and drones, were participants along with the mentor discussed their potential for the humanitarian action. With the completion of the course I was able to link causes and traits of humanitarian emergencies with the potential of geospatial monitoring capabilities. I understood opportunities and challenges of latest geospatial technology in humanitarian action.
As my seminar paper I presented on the topic, "EO*GI in Local Risk and Vulnerability Assessment".
With the state of art technologies, individuals have been able to collect, analyzing and using complex spatial data to help countries in dealing with the local and national challenges like natural hazards. With the use of Earth observation sensors to collect spatial data of the Earth's surface, which are then modelled with several geospatial tools to obtain information on the probability of damage and potential danger of the location or situation (mostly without physical presence). These tasks help decision-makers to implement efficient decision for further actions to be effective in saving potential damage to life and infrastructure.
The course dealed with the theory of cartographic communication, principles of map design, methods for preparing thematic maps, and examples for the application of thematic cartography in various subject areas. The course was designed around several projects. The projects cover selected topics of geovisualization and thematic cartography ranging from thematic mapping, to web-based mapping and additional forms of representation in cartography. The projects revolved around: revision of cartographic guidelines for thematic maps including tools in ArcGIS; Web-based mapping and interactive maps with GeoJSON; and Exploratory spatial data analysis.
With the completion of this course I was able to extent my knowledge and skills of thematic cartography. My project work included:
The course was dedicated to providing an introduction to data science and machine learning, leading to Artificial Intelligence. It included the concept of geo-spatial with artificial intelligence. The course covered the topic and the emergence of data science as a result of numerous trends colliding (internet of things, digitalization, open data, social media, etc.). It then evaluates the new data's qualities, as well as the data science and machine learning ecosystem's technological and methodological environment.
With this course, I learned how to start a data science project, resolve data analytics concerns, and apply data science to real-world problems through examples and instances. As an understanding of the basis of data science, I along with my team member Tanya Singh perform a task entitled: "Basic Statistical Analysis using RNN for hurrican OPHELIA.". Though we were not able to obtained accurate result, we were happy for being able to perform the task well and enhancing a deeper understanding of the topic. We have documented our scripts here.
This curriculum was introduced to us as a course dedicated to gaining an understanding of Application Development in an OBIA software environment using Cognition Network Language (CNL) (eCognition). Throughout the course, we learned about the CNL language and worked through a variety of applications using it. A wide range of topics related to Object Based Image Analysis were discussed. As part of the course, we worked in groups of two to create our own GUI as an eCognition Architect Solution. In my case, I chose a topic of "GUI development for Dwelling Extraction using Deep Learning approach in eCognition" with a team member based on the knowledge I gained during my internship.This document contains a summary of our work.
This course provided us with a thorough understanding of a variety of spatial analytical techniques as well as the skills needed to apply these techniques expertly to specific problems. I gained a better understanding of geospatial analysis techniques and am now more familiar with spatial relationships, interpolation, geostatistics, process modeling, and other related topics. Throughout the session, I submitted my work as ArcGIS Storymaps on a regular basis. They were created with the following title:
The course covered Object Based Image Analysis techniques in depth. Despite the fact that it was a self-paced course using freely available online materials, supervision was provided via online discussion. With the completion of this course, I gained skills in applying spatial concepts in image analysis by handling basic technical principles of image segmentation and object-based classification and validations.
The course focused on communicating established and upcoming architectures introducing the conceptual strategies, organizational requirements and legal frameworks for leveraging the advantages of Open GIS. It provided us with organizational, legal and technical foundation for accessing, using and delivering geographic information in harmonized spatially enabled distributed IT- service infrastructures.
Based on these concepts, I learned how to utilize open, shared GIS resources and developed my understanding and ability to design and use Open GIS data structures, workflows and processes leveraging open information repositories.
The course introduced concepts, research methodologies, and techniques of scientific writing with regards to (GI)Science. With this we were able to get an idea on how to choose the right media to publish results as well as details about the iterative process of peer-review in scientific journals from submitting the manuscript to the final article appearing in print. The whole course is accompanied by exercises in scientific dissemination in the English language.
With the completion of the course I was able to compose an academic paper entitled "Spatial Concept Matters in Land Administration" on scientific principles, along with an abstract followed by a final presentation.
The course included a review of statistical concepts as well as an introduction to spatial statistics and geostatistics. Spatial statistics presented methods of point pattern analysis such as nearest neighbor analysis and kernel density estimation, whereas geostatistics covered interpolation and estimation methods such as deterministic interpolation and kriging. In the course, the statistical software R was introduced, and ArcGIS was used for the practical sessions. With the completion of this course, I gained an understanding of statistical methods in the context of spatial data, as well as the ability to use R.
With the completion of this course, I have a good understanding of several machine learning methods and the actual mathematical concepts implemented behind them. I believe that I can now design and implement machine learning methods within a standard framework. This sub-module under artificial intelligence included chapters on: Principles of supervised learning and other ML paradigms; Classification and regression; Dimension reduction and feature selection; Anomaly detection, Training strategies and evaluation protocols and; Use of software libraries.
This course included the introduction to principles of neural networks, optimisation, regularisation and transfer; along with a few of the state-of-art architecture used for several applications in AI for Earth Observations.
It was introduced to us as a course that will help us understand the magic shown by EO software. With its competition, it fulfilled the objective in my case as I now have a better understanding of mathematical morphological tools, tress, and graph-based image processing to understand attribute profiles and pattern spectra and a concept of multiscale segmentation.
Global and local image feature analysis methods were presented during the course work. This module also introduced and helped to understand and implement several deep learning methods for image segmentation(instance and semantic) and object detection.
Itowns, Kepler, .....
As the workshop's task, we as a group of 8 joined an international contest hosted by Microsoft AI for Earth, "On Cloud N: Cloud Cover Detection Challenge". It immensely helped us put our course knowledge to use in a practical application. After completing the workshop, a combined report was presented that summarised methodology and concept adapted during the competition. A copy of the summarised report submitted by our group is here.
As a part of my course work, I got an opportunity to participate in spring school organized by the Interfaculty Dept. of Geoinformatics – Z_GIS, University of Salzburg. Spring school was held from 7th-9th April 2021 in online mode where.....Read More
I got an opportunity to participate in EO4GEO International Summer School: Intelligent Earth Observation 2021 co-organized the University of Salzburg and UNEP/GRID-Warsaw Centre, in the framework of the EO4GEO project. Summer school was held from 6th June-9th July 2021 in online mode where.....Read More
I also got an opportunity to participate in P.A.I.S.S in France.
From March 1, 2021 to July 24, 2021, I was hired as a paid part-time intern for Spatial Service GmbH for humanitarian related project. Due to confidential reason though I am not allowed to share images publicly, I am very much happy to share my experiences and learning while at work in Spatial Services. I have tried in summarize it in this document.