Transforming Agriculture: Drones Enhancing Livestock Management with AI Precision

In Colombia, livestock farming represents an important source of income, contributing 27% to agricultural Gross Domestic Product. However, this activity faces various challenges to ensure its proper development, such as cattle theft, known as rustling. This problem is what we seek to address with this project, which proposes a counting and monitoring system through aerial surveillance of livestock. Such results are achieved using images captured by a drone and further processed by an artificial intelligence (AI) algorithm. Remarkably, a key aspect of the project is the establishment of a database, collecting 1546 aerial images of livestock in 4 locations in Santander and Boyacá, Colombia. After rigorous labeling of the images, we proceeded to train the YoloV8m model. With the implementation of this model, we achieved an accuracy of 95.1%, precision of 96.1%, a sensitivity of 98.9%, and F1 metric of 97.47%. These results validate the model's ability to detect cows.