Using a system based on computer vision and machine learning, university students develop a unique approach to identifying and protecting diverse Lebanese flora
Located in a region where some 2,600 different vascular types of plants thrive, Lebanon is part of a Mediterranean “hot spot” known for its extensive diversity of plant species. And for a team of professors and students at the Maroun Semaan Faculty of Engineering and Architecture at the American University of Beirut in Lebanon, being able to identify and protect these species – especially given a global shortage in taxonomy experts — was critical. Thanks to their recent EPICS in IEEE project, “Mitigating the Taxonomic Impediment Problem of Plants Using ML and Citizen Science,” the team’s innovative development of a smartphone app using computer vision (CV) and machine learning (ML) is helping to map the many different types of flora throughout Lebanon and identify and protect any endangered species.
Left to right are Salma Talhouk (Professor, Landscape Design and Ecosystem Management), Dany Abou Jaoude (Assistant Professor, Mechanical Engineering), and Ibrahim Issa (Assistant Professor, Electrical and Computer Engineering).
In the following interview, faculty leads Dany Abou Jaoude (Assistant Professor, Mechanical Engineering), Salma Talhouk (Professor, Landscape Design and Ecosystem Management), and Ibrahim Issa (Assistant Professor, Electrical and Computer Engineering) discuss the goals of their project, which was conducted in partnership with Lebanon’s Al-Shouf Biosphere Reserve, as well as their hopes for the outcome and their positive experience with the EPICS in IEEE program.
Please discuss the scope of your project and the need for this effort within your local community.
Professor Talhouk: Monitoring plant diversity is key to understanding and responding to climate change and biodiversity loss. The current global shortage of taxonomic experts limits opportunities to monitor plant diversity on a large scale and regular basis, so our project aimed to leverage technology to address this problem in Lebanon. Given that our local community of biosphere reserve workers and volunteers lack scalable and user-friendly tools for data collection on existing plant species and their distributions, our objective was to create a tailored smartphone application using ML to classify plant species in Lebanon. Under the guidance of experts at the Shouf Biosphere Reserve, 110 student volunteers visited the reserve from May to September 2024 and captured images of spring, summer, and fall flowering plants to create a robust database of plant images.
To fill the gap in taxonomy experts, we tested multiple state-of-the-art ML models for use in classifying the collected data. With just a camera and information on the location and time, our app will be able to identify species names from their physical characteristics, geolocation, and season and serve as a real-time plant recognition tool that covers all species, areas, and seasons in Lebanon.
What challenges did you encounter on your project and how did you address them?
Professor Jaoude: Student enrollment in our project exceeded our expectations, so we had to organize our field trip and data collection processes for maximum efficiency and engagement. In addition, armed conflict in Lebanon during Fall 2024 brought field visits to a halt, which led to a slight delay in data cleaning towards the end of the project; happily, however, we were able to meet our target for plant species numbers.
What are the results of your project so far?
Professor Issa: In addition to the Shouf Biosphere Reserve team’s overwhelmingly positive response to our project idea and the important dataset collected, we’re very pleased by the engagement and impact the field trips have had on our volunteers. On the technical side, students gained key skills in machine learning, artificial intelligence, computer vision, data collection, storage, labeling and handling, GIS mapping, wireframe development, and benchmarking against plant identification apps. Students also got hands-on experience in the importance and difficulty of data collection for machine learning purposes, gained an appreciation for the usefulness and impact of machine learning advances on problems of local relevance, and became more aware of and sensitized to biodiversity issues overall.
What future activities will your project involve?
Professor Talhouk: There will be more field trips engaging even more student volunteers to collect additional data from different seasons as well as a crowning workshop that celebrates our project efforts and provides an overview of our project goals, machine learning aspects, and taxonomy and labeling issues for incoming student volunteers. Articles will also be published documenting our collected datasets, research findings, and machine learning models.
What’s your hope for your project and the role it will ultimately play in the community (or world)?
Professor Jaoude: We hope to repeat this pilot project in other areas of Lebanon and complete the colossal task of data collection on Lebanon’s 2,600 plant species in order to have a fully functional smartphone application that provides the needed information to users (eco-tourists, biosphere managers, etc.) and allows for the creation of spatial-temporal species distribution maps. We’d love to replicate our efforts in other Middle East and North Africa (MENA) countries as well as countries around the Mediterranean Sea. Given that it will be an impossible task for academic experts to complete on their own, our goal is to popularize taxonomic efforts by allowing citizen scientists to contribute.
Finally, what would you like to share about the support you received from EPICS in IEEE and the value of the opportunity to participate in an EPICS in IEEE project?
Professor Issa: Our experience with EPICS in IEEE has been great. EPICS in IEEE projects provide students with real-life examples and experience in how technology can empower society and contribute to nature conservation. The support we received from EPICS in IEEE was instrumental to our project by enabling our field trips and student volunteer engagement. We also received positive publicity through news articles about our efforts and by participating in a panel entitled “Artificial Intelligence and Advanced Technology in Service Learning” and sharing our insights from the project. Overall, EPICS in IEEE projects successfully leverage technology to solve real-world problems that affect our communities, and we strongly encourage participants to engage in these types of transdisciplinary academic activities and community partnerships to address local yet complex issues.
This project was made possible by $3,810.00 in funding from the Antennas and Propagation Society (APS), an EPICS in IEEE partner.
For More Information
For more information on EPICS in IEEE or the opportunity to participate in service-learning projects, visit https://epics.ieee.org/. “EPICS (Engineering Projects in Community Service) in IEEE” is an initiative which provides opportunities for students to work proactively with both engineering professionals, technological innovation, and local organizations/partners to develop solutions that address global community challenges.
Recent Comments