Food insecurity is a huge problem that is affecting countries under development, particularly in sub-Saharan Africa. The ability to collect data with high resolution and field wide in precision agriculture has come to the attention of key players in agronomic crop production as well as in agronomical research due to its high accuracy and efficiency compared to the traditional methods used to be popular over the past years. The most important objective of this research work is to establish a model that would enhance food security measures through the integrated use of Unmanned Aerial Vehicles (UAV) as an IoT system and transfer learning based on the Convolution Neural Network (CNN) model to classify and monitor the crop conditions for earlier decisions making when necessary.

The proposed system will put in place methods to monitor crop conditions while predicting the presence of Fall Army Worm (FAW) in crops for farmers and the government to act accordingly. In the following master thesis, an IoT-based UAV system is integrated with machine learning techniques in order to increase crop production and reduce hunger that has been found in some areas of the country. The use of UAV with elevated multispectral cameras for agricultural practices provides spatial, spectral, and ground data used for monitoring and analyzing crop conditions, for increased crop production.

The major purpose of this work was to provide and analyze data on FAW categorization and presence in maize crops using a transfer learning strategy based on an Inception V3 pre-trained model that had been fine-tuned. The effectiveness of the suggested model is assessed using a variety of numerical calculations.