Years: 2024
Role: Project Manager
Funder: Google
This project aims to develop a prediction system for the stages of chronic kidney disease (CKD) specifically adapted to the African context, particularly in Benin.
Years: 2024
Role: (South) Project Coordinator
Funder: ARES/Belgique
This applied research project in artificial intelligence and optimization aims to create new optimization models for the automatic planning of defenses in the universities of Benin, which have some challenging (soft) constraints.
Years: 2024
Role: Member
Funder: INRIA/France
This project focuses on the development of new pest control methods based on a crop monitoring network system using the Internet of Things (IoT). The application is focused on pest monitoring in sub-Saharan African farms.
Years: 2024
Role: Member
Funder: European Union
This project aims to strengthen the skills of young African scientists in precision agriculture through cross-regional academic mobility, in order to improve climate-smart agri-food systems.
Reference paper: Sorghum Yield Prediction using Machine Learning, J. G. N. Zannou, V. R. Houndji. International Conference on Bio-engineering for Smart Technologies (BioSMART'2019), Paris, France.
Agriculture is one of our greatest assets. Estimation of a future agricultural production is an important challenge for farmers. In this work, we propose a system based on artificial intelligence to estimate farm yields. We use drones and Artificial Intelligence (in particular Machine Learning algorithms) to estimate an agricultural yield production. For the first experiments, we focus on the Sorghum. Sorghum is the fifth largest cereal with regards to volume of production, after maize, rice, wheat and barley. It is the main cereal for many low-income populations living in the semi-arid tropics of Africa and Asia. We propose and experiment an approach with Machine Learning to estimate the agricultural yield of a farmland of Sorghum before the harvest. These algorithms allow us 1) to detect the different ears of Sorghum on an image and 2) to estimate their weight. On our dataset, we obtain an average accuracy of 74,5% for the detection of sorghum.
Early and accurate diagnosis of plant diseases is a vital step in the crop protection system. In traditional practices, identification is performed either by visual observation or by testing in laboratory. The visual observation requires expertise and it may vary subject to an individual which may lead to an error while the laboratory test is time consuming and may not be able to provide the results in time. To overcome these issues, some image based machine learning approaches to detect and classify plant diseases has been proposed. In this work, we focus specifically on the identification of healthy plant and three types of rice plant diseases (Oryza sativa) namely Bacterial Leaf Blight (BLB), Brown Spot (BS), Healthy leave (H), Leaf Smut (LS).
Reference paper: AmonAI : a students academic performances prediction system, I. Houndayi, V. R. Houndji, P-J Zohou, E. C. Ezin. Accepted at the 11th EAI International Conference on e‐Infrastructure and e‐Services for Developing Countries.
To ensure a great training of the students, it is important for them to receive adequate support to improve and succeed in their studies. Unfortunately several conditions, the highest number of students mainly, make more difficult to monitor students. To reduce the failure rate of students we iniate this project. It will then dive into the application of machine learning techniques on the available data to make the prediction of academic performances of the students in the LMD system. We work on AmonAI, a system that allows to anticipate their results in order to reduce their failure as much as possible by taking appropriate decision. The system makes the prediction of the students academic performances through classification and regression in each teaching unit of a semester and provide visualizations based on these predictions.
In the health sector, practice is very important for surgeons who must master the surgical procedures as well as the techniques and tools used for the success of each surgical operation. In this work, we realized an accessible tool that can help them to carry out sessions of practical work in an ideal virtual environment, by putting at their disposal a medical simulation application in virtual reality. We propose in this work an implementation of a virtual reality simulation module of laparoscopic appendectomy. The prototype is totally functional.
In this work, we have produced an accessible tool that allows students to complete and reinforce the learning of measurements and observations. Called ArGeo, this application offers some immersive experiences using Augmented Reality. We propose the implementation of a module on geology at the secondary level. Various tests were carried out and show that the implemented module is functional.
Nowadays, security is central, whether protecting property or ensuring individuals' safety. Although initially designed to address these needs, traditional surveillance systems exhibit notable shortcomings. Personal security, required to monitor screens constantly, may occasionally overlook crucial sequences, jeopardizing security. Determining when to record camera feeds can be intricate, and making these recorded sequences actionable presents an additional challenge. It is in this context that NextVision emerges. Our project endeavors to harness the potential of Artificial Intelligence (AI) to revolutionize video surveillance. Positioned as an intelligent and user-friendly video surveillance platform, it relies on a skillful fusion of computer vision and natural language processing for real-time automatic object detection. Supported by a Django-based backend architecture and a Next JS-powered frontend, synchronized in real-time via sockets, NextVision embodies a continuous evolution powered by machine learning to enhance security for all proactively.
Guided tours in the tourist sites of Benin are generally always classic with a human guide who tells the stories related to the tourist places visited. To enhance the visitor experience, we offer an augmented reality-based system that allows the visitor to ex- perience the story by observing it directly in 3D while the guide exposes the stories. We thus offer visitors an attractive experience in the hope of making the slave route a popular destination. The experiments show that the prototype ready.
...
In the health sector, practice is very important for surgeons who must master the surgical procedures as well as the techniques and tools used for the success of each surgical operation. In this work, we realized an accessible tool that can help them to carry out sessions of practical work in an ideal virtual environment, by putting at their disposal a medical simulation application in virtual reality. We propose in this work an implementation of a virtual reality simulation module of laparoscopic appendectomy. The prototype is totally functional.
We propose a system that predicts the purchase intention of users visiting e-commerce platforms, from data collected when they navigate on these platforms.
...