Omdena Chapter Page: Bangladesh

Omdena Bangladesh Chapter - Omdena Chapters

Welcome to the Bangladesh Local Chapter!

Apply here to be a chapter lead for other cities and/or universities in Bangladesh.

Upcoming Projects

Project Applications Open!

Project Starts: September 9th, 2022 

Duration: 4 Weeks

Fertilizer Recommendation System to Ensure High Crop Yields in Bangladesh
Bangladesh Chapter Lead – Kazi Saiful Islam Shawon

 

The problem

According to a recent study by the UN’s Food and Agriculture Organization about 520 million people are without food security. The pandemic worsened access to food for these large numbers of people. As much as 31.9 percent of the population in Bangladesh experienced moderate to severe food security. As a direct consequence of rapid growth in population and urbanization, the situation is deteriorating rapidly. It is assumed the land will be reduced by 5% in 2030 and by 8% in 2050. The research shows that the total demand for rice in 2030 and 2050 under the BAU (Bangladesh Bureau of Statistics) scenario would be 40.11 million tons and 46.15 million tons. The total direct demand of rice for consumption in 2030 will be 32.4 million metric tons (MMT), which is a 14% increase from the 2015 level. As for that, the share of rice will decrease from 82% in 2010 to 79% in 2030 and 78.6% in 2050.

There is a pressing need and opportunity to improve food security by enacting effective and coordinated Artificial Intelligence-driven (AI) solutions and actions, which will necessitate significant improvements in the relevant sectors, such as crop classification, fertilizer recommendation, crop productivity, etc.

 

The solution

 

To ensure food security the most important aspect arises is high crop yield. The goal of this project is to utilize remote sensing data for building a fertiliser recommendation system targeting crop productivity. The objective of this study is to analyze certain crops such as rice, wheat, etc. of various regions in Bangladesh, and measurement of the fertilization uptake for high yield.

The project aims to deliver a data-driven solution for fertilizer recommendation of the interested crop fields in Bangladesh. This recommendation system will be live through web app where users will get fertilizer suggestions of their interested fields. The results will be made open source. This will help farmers to increase crop productivity.

 

The project goals

The goal of the project is to

  • Collect data from open source satellite images of Bangladesh and extract necessary information with remote sensing analysis.
  • Process the data following a systematic methodology, and do exploratory data analysis of harvested crops.
  • Develop an automatic recommendation system for appropriate fertilization requirements of certain crops.
  • Outline an AI-driven solution to build a system to improve crop yield for the farmers.
  • A web app where users would get their field statistics and recommendation for their crops.

 

The learning outcomes

  • Remote Sense Analysis.
  • Data and GIS Preprocessing.
  • Data Visualizations.
  • Machine Learning.
  • Web app development with Georeference API development.
  • Real-world impact to improve the agriculture sector.

 

The project timeline

Week 1 Week 2 Week 3 Week 4

– Data Collection

– Brainstorming

– GIS Processing

– Remote sense analysis

– Exploratory Data Analysis(EDA)

– ML Preprocessing

– Pre-processing Completion. 

– Building Machine Learning models.

– Evaluation of models on certain regions.

-Building Recommendation System

– Creation of maps of certain crops across Bangladesh.

– Georeferencing the classified maps.

– Web app for fertilizer recommendation.

 

  •  

 

Previous Successful Projects:

 

Ongoing Projects

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

Project Applications Closed!

Project Started: April 16th, 2022 

Duration: 4 Weeks

 

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.

 

The solution

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 project goals

The goal of the project is to:

  • Collect data from open-source satellite images of specific rivers of Bangladesh, such as Buriganga, Meghna, 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 and deliver geolocations of them.
  • Outline an AI-driven solution to improve the cautioning system for Bangladesh.

 

The learning outcomes

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

 

The tasks & timeline

 

Week 1 Week 2 Week 3 Week 4

– Data Collection of certain rivers.

– Brainstorming

– GIS Processing

– Remote sense analysis

– Exploratory Data Analysis(EDA)

– ML Preprocessing

– Pre-processing Completion. 

– Building Machine Learning models or computer vision models for illegal occupation of rivers.

– Evaluation of models on certain regions.

– Creation of maps of certain encroachment of Bangladesh rivers.

– Georeferencing the classified maps.

– Deployment of the model to classify encroachment with georeferenced.

 

The learning outcomes:

1. Remote sensing analysis.

2. GIS Processing.

3. Machine Learning Classification.

4. Computer Vision detection.

5. Collaborative Project Management.

6. Leadership.

 

Completed Projects

AI for Improving Road Safety in Bangladesh

Project Applications Closed!

Project Started: June 26th 

Duration: 4 Weeks

 

The Background

Road safety is a major concern in Bangladesh. An estimate states that 55 people are killed in road crashes every day, and that vulnerable road users including walkers, motorcyclists, and unsafe and informal public transportation users account for more than 80% of road traffic deaths [1] . As a direct consequence of rapid growth in population, motorization and urbanization, the situation is deteriorating rapidly. The potential of maturing Artificial Intelligence (AI) and Internet-of-Things (IoT) technologies to enable rapid improvements in road safety has been largely overlooked. There is a pressing need and opportunity to improve road safety by enacting effective and coordinated Artificial Intelligence-driven (AI) policies and actions, which will necessitate significant improvements in the relevant sectors, such as better enforcement, better roads including improving design to eliminate accident black spots, and improved public education programs.

The Problem

The first step would be to identify the root cause of road accidents in Bangladesh by analyzing data, and then investigate the utility of AI driven technologies i.e. automated analysis of traffic scenarios,  monitoring speed, violations and driving patterns of the drivers, auto-alerting drowsy drivers etc. This will help to outline a solution tailored for the infrastructure of Bangladesh.

 

The project results will be made open source. The aim being to help ride sharing service providers, regulatory bodies, policy makers etc. while educating aspiring data scientists in solving real-world problems.

The Project Goals

  1. 1. Collect road accident-related data from public databases, newspapers, web pages, etc.
  2. 2. Analyze the data following a systematic methodology
  3. 3. Do exploratory data analysis
  4. 4. Outline an AI-driven solution to improve road safety.

The Learning Outcomes

  1. 1. Web Scraping
  2. 2. Data Cleaning
  3. 3. Natural Language Processing
  4. 4. Machine Learning
  5. 5. Developing Dashboards

Source Code:  https://github.com/OmdenaAI/omdena-bangladesh-roadsafety

Demo:

Developing an AI-driven & Secure Online Marketplace
Project Started: October 9th 2021
Duration: 4 Weeks

The Background

Retail e-commerce in Bangladesh is growing at a rate of 72 percent every month. This industry currently employs over 35,000 people and over 25,000 small and medium businesses (SMEs). Up to 2018, there were 2,500 e-commerce business sites and 150,000 e-commerce pages, respectively. Currently, the number of deliveries per day is believed to be between 15,000 and 20,000 [1]. Despite the fact that the e-commerce business has grown over time, it still faces a number of challenges including a shortage of data science and machine learning driven platforms, enterprise applications powered by AI, tools for developing conversation interfaces, etc. Further obstacles include a lack of trust in the e-commerce ecosystem, product quality control and a fear of shopping online. Another important flaw is the absence of a proper procedure for resolving customer complaints. Moreover, there is a reluctance to incorporate banking facilities in the e-commerce industry [2]. As per Statista, Bangladesh is in the 47th position in the world of e-commerce and the share of e-commerce sales is a small percentage of total retail sales of the country.

 

The Problem

The objective of this study is to analyze the online market places available in Bangladesh e.g. daraz, bikroy.com, chandal, foodpanda, etc. to analyze the currently offered technologies, products, compare prices with physical stores, analyze user reviews and sentiments, and identify the shortcomings related to technology, trust, financial transactions, and business policy.

The project aims to deliver a data-driven solution for improving the market places including fraud detection/prevention in the context of Bangladesh. The results will be made open source. This will help regulatory bodies, policy makers, e-commerce companies, consumers etc. and also educate aspiring data scientists in solving real-world problems.

 

The Project Goals

1. Collect data from online market places in Bangladesh related to product, technology in use, user-reviews, complaints, survey-reports etc.

2. Process the data following a systematic methodology, and do exploratory data analysis

3. Develop an AI-driven product and deal recommendation system.outline an AI-driven solution to improve the online marketplace eco-system in Bangladesh.

 

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4

– Web Scraping

-Market Survey

-Data Collection

– NLP pre-processing

– Exploratory Data Analysis(EDA)

-Interactive plots with Real-time data

-Sentiment Analysis Model 

-Recommendation System

-Fraud detection system

-Deployment of recommendation system

-Deployment of fraud detection system

summarize a comprehensive business insights from the study

Register for future updates!

Bangladesh Chapter Lead

Kazi Saiful Islam Shawon

Kazi Saiful Islam Shawon is a Deep Learning, Computer Vision enthusiast, and ML Engineer at Spacenus GmbH. He is teaching computer vision courses on various platforms. At Omdena he has worked with various impact startups to develop computer vision applications. Shawon is particularly interested in applying artificial intelligence and computer vision to solve real-world problems focusing on ecological benefits, economic enhancement, and sustainability for a better global society

Our Partner(s):