Omdena Chapter Page: France

Omdena France Chapter - Omdena Chapters

Welcome to the Omdena France Chapter!

Upcoming project(s)

Duration: 19th February – 19th March
Real World AI project!
Improve sorting and segregation of waste using machine learning
Chapter lead for the France chapter: Rebecca Alexander

The Objective

Development of image recognition techniques to improve sorting and segregation process of solid waste management.


The Background

Solid Waste Management is a universal issue, and at the same time, it is the ‘need of the hour’ project. One of the major contributors to municipal waste is plastic waste and its generation and consumption have increased drastically over the past few years even without developing a strategy to manage the waste generated.  UNEP reports suggest that so far, only 9% of all plastic waste produced after the 1950s was recycled, and the rest ended up either in landfills or in our environment [1]. The current recycling rate, improper management of the generated waste, and its accumulation in the environment pose a massive threat to the marine and land habitat. Studies indicate that even the remote areas of ocean and land ecosystems are affected by the scourge of plastic trash, chemicals and other pollutants. One such example is the Great Pacific Garbage Patch, a marine debris collection spot in the Pacific Ocean where several thousands of tonnes of ocean plastic are estimated to be floating on the surface [2]. The concern is not only about plastic waste: it is about all the trash generated: metallic, e-waste, organic, textiles etc.  We need to adopt different strategies and recovery plans to manage these waste materials in order to reduce the impact they caused on our ecosystem.  

Project Outline

The biggest challenge in recycling/re-using waste is sorting and segregating different types of waste since segregation of waste aids in targeted recycling or even decomposition. As an example, segregating a dry metal can from a metal can containing organic matter eases recycling. The necessary action for proper segregation of the waste on large scale is to identify various materials first. Once identified, neuromorphic tools could be used to sort things based on the identified parameters. However, while there exist several methods to identify different materials such as visual sensors, olfactory sensors as well as spectroscopic tools, there are very few or no attempts at using artificial intelligence to specifically identify materials from the waste, which could then be applied to ease the segregation process. We, therefore, propose to use visual image recognition to first identify objects, in their full form or by parts in order to be used later for segregation. For example, we envisage the identification of different materials such as plastics, metal and paper in a used milk carton, which could lead to proper recycling of plastics, paper and metal.



The Learning Outcomes

  • In this project, participants will be guided to perform the following steps: 

    1. Data Collection through web scraping and creation of image library 
    2. Image Preprocessing for Computer vision 
    3. Annotating Images to reflect the correct waste category 
    4. Computer Vision techniques to identify and classify different waste materials 
    5. Deploying Dashboard and Visualization to make the ML model available to the public 


The Tasks & Timeline


Week 1 Week 2 Week 3 Week 4


  • Find relevant data sources on correct waste segregation in France 
  • Image preprocessing & Annotations (as required)
  • Exploratory Data Analysis (EDA) 

  •  Image Annotations  

  • Debug and process videos/images for the training of ML models to detect correct waste categories. 

  • Image Classification model based on the waste category 


  • Start building Streamlit WebApp (with Tableau plots, if needed)



  • Finish Integrating WebApp 

  • Deploy the App in Cloud Application Platform


Collaborating researchers: Surya Abhishek Singaraju, Jennifer Joseph 



Completed Project(s)

Analyzing Impact of Lockdowns on Air Pollution in France

The Background

We regularly hear that transport is a major cause of air pollution. The lockdowns have drastically limited population movements by car, train or airplane. The idea is to see if these limitations have had an impact on air quality in France and in which proportions.

The Problem

Measure the impact of reduced mobility on air quality.

The Project Goals

Generating an Analysis Report on the affects of lockdown on Air Pollution






Project 2

Building a Tracking dashboard for Flu

The Background

Every year, flu has a significant impact on hospitalizations. Flu mortality highlights the seriousness of the disease and the importance of vaccination for people at risk. Additionally, barrier measures are indispensable to limit the spread of the virus by direct contact.

Like Covid tracking tools, a flu tracker can make this information accessible.

The Project Goals

The aim is to use the available open-source data on infections, deaths and vaccinations, in a way similar to what is done for Covid-19. This will inform people about public health and best practices in a simple way.

The project results will be made open-source. This project will also reveal the severity of influenza compared to Covid.

The Tasks & Timeline

Week 1 Week 2 Week 3 Week 4

Find relevant data sources

– Developing web scraping scripts, if needed

– Exploratory Data Analysis(EDA)

-Interactive plots with Real-time data

-Interactive map vaccination tracker 

– Start building Streamlit WebApp (with Tableau plots, if needed)

-Finish Integrating WebApp

-Deploy the App in Cloud Application Platforms

The Learning Outcomes

1. Scrape and parse open data

2. Exploratory data analysis to get main insights

3. Data modelling (for example model the risk of an event with nb of people and other variables)

4. Build a dashboard using data visualization tools using Streamlit and Tableau


Omdena France Chapter
France Chapter Lead

Rebecca Alexander

Rebecca is a Physicist-turned-Data-Science professional, who enjoys coding to decode patterns in data. Prior to this, she completed her Ph.D. at the French Atomic Energy Commission (C.E.A., Saclay). She’s always interested in debates about causality v/s correlation and Data Science projects.