Automatic River Encroachment Detection in Bangladesh with Machine Learning and Remote Sensing

Local Chapter Bangladesh Chapter

Coordinated byBangladesh ,

Kazi Saiful Islam Shawon

Status: Completed

Project Duration: 16 Apr 2022 - 23 May 2022

Open Source resources available from this project

Github Repository

Project background.

Satellite imagery promises to improve this issue in bangladesh. With the open source satellite data an illegal encroachment can be detected with remote sensing data analysis, GIS processing , illegally filling up detection on river banks and illegal occupation for infrastructure detection on river banks.

The problem.

River encroachment is a threatening problem that has been going on for a few years now. Bangladesh Govt. has been relentlessly working to free rivers from encroachment after ensuring their navigability by dredging them round the year. BIWTA has removed a total of 18,025 illegal structures from riverbanks between 2010 and 2019 and reclaimed 738 acres of riverbanks from encroachments in Dhaka and Narayanganj river port areas. The National River Conservation Commission says 50 percent of the rivers have been freed from the clutch of grabbers, but the rate is only 7 percent in Barishal, the lawyer said. Bangladesh has around 800 rivers. The State Minister recently said over 65,000 grabbers have encroached on the rivers.

Project goals.

The goal of the project is to:

- Collect data from open-source satellite images of specific rivers of Bangladesh, such as the Buriganga River, Meghna River, etc, and extract necessary information with remote sensing analysis.
- Process the data following a systematic methodology, and do exploratory data analysis of illegally occupied river banks.
- Develop an automatic encroachment detection for certain rivers with machine learning and deliver geolocations of them.
- Outline an AI-driven solution to improve the cautioning system for Bangladesh.

Project plan.

  • Week 1

    – Data Collection of certain rivers.
    – Brainstorming
    – GIS Processing

  • Week 2

    – Remote sense analysis
    – Exploratory Data Analysis (EDA)
    – ML Preprocessing

  • Week 3

    – Pre-processing Completion.
    – Building Machine Learning models or computer vision models for illegal occupation of rivers.
    – Evaluation of models on certain regions.

  • Week 4

    – Creation of maps of certain encroachment of Bangladesh rivers.
    – Georeferencing the classified maps.
    – Deployment of the model to classify encroachment with georeferenced.

Learning outcomes.

– GIS processing.
– Remote Sense Analysis.
– Data Preprocessing.
– Data Visualizations.
– Machine Learning.
– Georeference API development.

Share project on: