DATABIO PROJECT: BETTER CONTROL OF WEEDS THANKS TO ARTIFICIAL INTELLIGENCE IN ORGANIC FIELD CROPS

Agri-Fusion has initiated a large-scale project, named DataBIO, with the Institute for Research and Development in Agro-Environment (IRDA) and Agrisoft to fill a need for expertise in the treatment of information and analysis of a large volume of data. Currently, nearly 3,000 hectares are dedicated to organic plant production within the company. Many data concerning crop operations are generated from different sources without their potential being fully exploited. The automation of data collection and processing will ultimately optimize the monitoring of farm operations, help decision-making and produce records. A pilot project is underway at the Agri-Fusion farm, in which a structured and standardized database has been developed to allow the statistical processing of variables useful for decision support at the field level. for the period from 2010 to 202. These variables were identified through a literature review and expert consultation. Data analysis will determine which variables have the most weight on weed control. However, some variables, although they have a recognized impact on agricultural yields, are not systematically monitored in agricultural businesses. This results in a risk of error in the diagnosis of the causes that influenced yields and thus, non-optimal decision-making. According to Evert et al. (2017), yield losses caused by weeds, estimated at 32%, are greater than those caused by insects (18%) and diseases (15%). In organic field crop production, weed control is recognized as a determining factor in crop productivity, in addition to representing significant labor and machinery costs. Weed suppression is generally done mechanically and must be carried out within an often very short time frame. In fact, specialized equipment can be used at specific stages of the main crop and weeds and in climatic conditions allowing the passage of machinery. This problem was highlighted by Agri-Fusion representatives since optimal yields cannot be obtained in certain fields due to an excessive proliferation of weeds that can lead to the abandonment of the crop. This project aims to introduce variables related to weed monitoring in the database, in order to consider this determining variable in the productivity of organic field crops. Using ML and AI will help prioritize weed control interventions based on crop stages and weather forecasts.