Omdena Chapter Page: Egypt

Omdena Egypt Chapter - Omdena Chapters

Welcome to the Egypt Chapters!

There are 3 active chapters in Egypt:

  1. Alexandria, Egypt
  2. Cairo, Egypt
  3. Giza, Egypt

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

Upcoming Projects!

We will be running an AI project soon…. Stay Tuned!

Ongoing / Completed Projects

Using data science for lung disease detection

The Background 

In partnership with GDSC, Egypt-Japan University of Science and Technology, I agreed with the team to run a project to get college students introduced to data science. Thus, the aim is to educate students on:

  1. Collecting datasets
  2. Modelling
  3. Web deployment

The Problem

In this 4-weeks project, the goal is to expose students to building models for lung disease detection. They will start with data collection all the way through deployment.

 

The Project Goals

Partntership goals:

  1. Promote Omdena through GDSC
  2. Network with other communities

Lung disease detection

  1. Check whether a CT scan of lungs to check the existence of disease
  2. Semantic segmentation of the disease
  3. web deployment

 

The Learning Outcomes

  1. Data Pre-processing
  2. Supervised learning models and Unsupervised learning models
  3. Interactive Graphs and Dashboard.

 

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4

  • Data collection
  • Research
  • Workshops

 

  • Data collection
  • Modelling
  • Web design

 

  • Modelling

  • Web deployment

  • Final deliverables
  • Reflection
Using AI-based models to detect plant disease

The Background 

Agriculture is one of the main pillars of some economies, especially the developing ones. However, some farms are distant, and there are no botanists to take care of the plants in a proper way. Therefore, some crops are wasted because of the lack of care provided fr the plants. Also, some farmers do not have a solid background on how to detect and plan emergency plans to protect the crops

The Problem

Due to the scarcity of financial resources available to some farmers in some developing countries like Egypt, they cannot afford to have botanists on their farms to take care of the crops; which leads to a situation in which the crops to be wasted. Hence, this project aims at building AI-based models to detect potential plant diseases and provide some emergency plans to contain the damage..

 

The Project Goals

      1. Collecting a plant disease dataset
      2. Data Pre-processing and Exploratory Analysis
      3. Building AI-based classification models
      4. Building emergency plans recommender system to contain the damage
      5. Deployment of the models to be used by farmers in distant and unfortunate ares

       

     

The Learning Outcomes

      1. Data Collection
      1. Data preprocessing
      1. Modeling
      1. Recommender systems
      1. Deployment

     

 

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4

Data collection and annotation

-Recommender system exploration

-Image pre-processing

 Exploratory Data Analysis (EDA)

  • Data collection and annotation

– Modeling

  • building an  initial recommender system 

-Budiling deployment frontEnd and basic API functions

-Modeling and voting on the best model

-Recommender system delivery

-Integrating the API functions with the models

-Integrating the models to be used through APIs along with the recommender system
Using deep learning for skin cancer detection

The Background 

According to  the partnership agreement that took place between Omdena, Alexandria, Egypt chapter and Google Developers’ Club of Alexandria University, we would like to launch this project to introduce the community of the university to Omdena challenges.

The Problem

In this 4-weeks project, the goal is to educate the students of Alexandria university to build and deploy deep learning projects through an Omdena challenge. The team is supposed to use deep learning for skin cancer detection. They will start the process from a scratch as most of them are novice to this field. Thus, they will be guided through the whole process by the chapter lead.

 

The Project Goals

1.Promoting Omdena at Alexandria University

2. Providing a hands-on experience for college students on AI

3. Helping students acquire the skills needed for:

a. Data collection

b. Modelling

c. Deployment

The Learning Outcomes

  1. Data Pre-processing

 

  1. Deep learning with Python

 

  1. Supervised learning models and Unsupervised learning models

 

  1. Deployment

 

 

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4
  • Data collection
  • Annotation
  • Modelling
  • Data Collection
  • Modelling
  • Deployment
  • Deployment

     

  • Project presentation
Creating & Analysing Open Data About Egypt

The Background 

Egypt has a very active community of data scientists and students eager to learn data science. This community would naturally hold the most interest in analyzing datasets about Egypt both for research and for open-source projects. There is, however, few datasets directly about or related to Egypt or the Egyptian population.

The Problem

Collect, organise, and analyse different datasets directly about or related to Egypt.

The results should serve as a starting point for anyone searching for datasets about Egypt. Providing dataset sources, visualisations, and reports that potentially use multiple datasets to tell a story with data. The results should also create a new dataset for Egypt and encourage the creation of more datasets by the community.

 

The Project Goals

    • Create a comprehensive index of open datasets about Egypt.
    • Integrate, analyze, and visualize diverse open datasets about Egypt.
    • Create a new dataset related to Egypt.
    • Host results on a website

     

The Learning Outcomes

    • Learning how to search for, categorize and visualize datasets.
    • Learning how to use multiple datasets in one report to tell stories with data.
    • Learning how to create new datasets.

     

 

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4

-Determine key websites that list datasets about Egypt

-Decide how to organize datasets, visualizations, and reports

-Decide on a new dataset to create

-Start creating an index of datasets and websites that host them

-Start creating visualizations and reports using the datasets in the index

-Start scraping data for the new dataset

-Continue improving the index and creating visualisations 

-Start looking into ways to host the results

-Organise new dataset and label it if necessary

-Finalize index and make sure it’s well-organized

-Host visualizations and reports on a website

-Finalize new dataset, host it, and test it

Completed Project(s)

1. Building Open Source NLP Libraries & Tools for the Arabic Language

The problem

1. Arabic is the 5th most spoken language in the world and the 1st language of the Arab world countries, making it extremely important worldwide.

2. Arabic is grammatically complex and has free order properties, which all pose significant challenges in Arabic NLP applications.

3. There are 3 types that characterize Arabic, including Classical Arabic, Modern Standard Arabic & Dialect Arabic.

4. Tools built by big tech and accessible to the majority of the world are limited to translating only a few of the most popular languages.

The Project Outcomes

The envisioned deliverables can be broken down into two main areas:

1. Build open-source Arabic NLP libraries for sentiment analysis, morphological modeling, dialect identification, and named entity recognition

2. Build 5:8 core functions to support Arabic NLP (lemmatization, stop words, tokenizing text, word embedding, part of speech tagging.. etc.) like NLTK but for Modern Standard Arabic.

Source Code: https://github.com/OmdenaAI/Arabic-Chapter

Demo: https://www.youtube.com/watch?v=PaWCX2IG7eo

2. AI Techniques to Reduce the Wasting of Freshwater

Despite the government’s efforts to save every drop of water as the country faces water scarcity, 98.4 million Egyptians still live under the water poverty line by 50 percent, below the international line of 1,000 m3. The danger of the water crisis in Egypt increased with the presence of regional conflicts over the water of the Nile River.

Flood irrigation represents the biggest challenge to wasting water, as 77% of the Nile’s water is consumed through it.

 

The Project Goals

1. Innovation in finding solutions that reduce water consumption.

2. Collect data regarding water consumption.

3. Study irrigation habits’ effect on water security.

Source Code: https://github.com/OmdenaAI/omdena-egypt-freshwater

DashBoard: http://omdena-egypt.herokuapp.com/

 

 

 

Challenge 2

Finding Paths to Safety Following Natural Disasters with Satellite Imaging and AI

The Background 

Natural Disasters are problems in Japan, with risk of earthquakes, floods and tsunamis. Japan has well-developed disaster response systems, but densely populated cities and narrow roads make managing the response difficult. By giving individuals information about the safest ways from their homes and places of work, it will increase their awareness of the surrounding area and improve their preparedness.

The Problem

 Design a model collecting data about the local roads from satellite images, classify them and indicate the safest route to be taken from point A to point B. Design an interactive dashboard to display the safest route in a map.

By making individuals aware, it will improve their preparedness and it can be used within families to prepare disaster response plans, depending on their circumstances. To be used by individuals, families and groups, and foreign residents who may not understand local information. Further development will be covering more geographical areas and publicising on a local level.

 

The Project Goals

  • – collect satellite images and identify road characteristics 

    – build a model for scoring the roads in terms of their suitability for use in emergency

    – build a pathfinding model from A to B, combining it with road characteristics 

    – suggest safest path from A to B

    – publish interactive dashboards to display road characteristics and safest paths

    – arrange demonstration and publicise to local audiences

The Learning Outcomes

  • Learn how to:

    – Extract, process and classify satellite images

  • Work with OpenStreetMap
  • – Build a scoring model
  • – Apply pathfinding models
  • – Build interactive dashboards (Streamlit)

 

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4

– Collecting data about Arabic data science content

-Researching Neural Machine Translation and selecting a model architecture.

-Exploring keyword extraction for technical terms.

– Exploratory Data Analysis(EDA)

-Road Scoring Model

-Tutorial (Streamlit)

-Dashboard building

-Road Scoring Model 

-Pathfinding algorithms

–Dashboard building

– Integrating work

-Deploy dashboard app

 

 

Alexandria, Egypt Chapter

Project Starts: October 20

Duration: 4 Weeks

All Data Science Skills Welcome

Alexandria, Egypt Chapter

Project Starts: October 20

Duration: 4 Weeks

All Data Science Skills Welcome

Alexandria, Egypt Chapter: Building Deep learning models to detect suicidal posts in modern standard Arabic

The Background 

After the COVID-19 pandemic, many individuals started spending more time alone and online, leading them to start expressing their ideas through social media posts. In addition, several people lost their jobs leaving them with negative thoughts. The negative thoughts can quickly lead to suicidal thoughts.

The Problem

This project aims to build deep learning models to detect suicidal language in modern standard Arabic.

Due to the effects of COVID-19, negative ideas and impacts are increasing drastically. Moreover, a lot of individuals started to have suicidal thoughts due to the burden put on their shoulders. This project aims at detecting suicidal language to detect suicidal people in order to offer help to those who need it the most.

The Project Goals

1. Collecting modern standard Arabic suicidal posts dataset

2. Building deep learning models to detect the suicidal language

3. Web deployment

The Learning Outcomes

1. Data collection

2. Data cleaning

3. Modelling

4. Web deployment

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4

Dataset collection and Preprocessing

Data labelling

Modelling

Getting initial results

Sentiment Analysis Model

Web app for the model

Web deployment
Omdena Egypt
Egypt Chapter Leads

Mohamed Mohey

Giza, Egypt Chapter Lead

Mohamed Mohey is an ML engineer with great interest in using MLOps to help solve real world problems. From data collection, labeling, and analysis to model training, testing, deployment, and monitoring, all aspects of the ML pipeline are of great interest to Mohamed. He is also passionate about learning and helping others learn and share information.

Lilian Ugwu

A data scientist who has an interest in using my skills to derive innovative social good, especially in the field of Healthcare.

Our Partner(s) :