The Fitri department, located in northern Chad, is in a region where agropastoralism is the dominant economic activity. However, in this region, the mobility of pastoralists is hindered by the degradation of the environment because of climate change and the fragmentation of agricultural land, which leads to recurrent conflicts between herders and farmers. This study proposes innovative methods based on remote sensing, artificial intelligence techniques (Deep Learning), and WebGIS for the automatic mapping and quasi-real-time monitoring of agropastoral resources in the Fitri department. The goal is to understand the evolution of pastoral areas and to optimize the prediction of pastoral mobility through a WebGIS model to reduce the occurrence of conflicts. The exploitation of Sentinel-2A multispectral images is done by applying convolutional neural networks (CNNs). The model was first trained on the training data, and then a predictive model was deployed for the automatic extraction of pastoral resources in quasi-real time. On the other hand, the cartographic database (BDcarto) was developed and implemented using geographic information layers under PostgreSQL and published on OpenLayer from GeoServer. The developed DL model and BDcarto are integrated into a WebGIS (pastoral guide) for the visualization and optimization of pastoral activities.