At UniLaSalle Amiens, a group of engineering students is tackling a very real challenge in the agricultural sector: automating the counting and analysis of potatoes during the sorting phases on industrial grading machines.
The project is being carried out in collaboration with ARVALIS – Institut du végétal, a national applied research organization that supports the agricultural sector and conducts numerous agronomic trials each year.
From prototype to artificial intelligence
The first step was to develop a vision-based counting prototype. This technical and software platform is already in use on sorting lines.
The students are now working to improve its performance using artificial intelligence-based detection and counting methods.
The goal is to increase accuracy, speed, and reliability, particularly for heterogeneous batches.
Concrete responses to field requirements
Current developments aim to:
- Improve the reliability of automatic counting
- Measure certain characteristics of tubers
- Explore the detection of damaged potatoes
- Adapt the device to different industrial equipment
- Simplify data exploitation (particularly in Excel)
The challenge is twofold: to reduce the drudgery of manual counting and to improve the quality of data collected during agronomic trials.
Through this project, engineering students are applying their skills in machine vision, image processing, artificial intelligence, and data analysis to address a concrete problem encountered by those working in the agricultural sector.
This is an example of collaboration between an engineering school and a technical institute to advance practices and support transitions in the sector.
