Ort: Hanse Innovation Campus Lübeck, Technische Hochschule Raum 02-1.02
Vortrag im Rahmen des Seminars „Künstliche Intelligenz in Anwendungen“, Externe sind herzlich willkommen
Dr. Nadeem Iqbal Kajla, MSN University of Agriculture Multan
In present centuries, the agricultural industry has witnessed remarkable progress driven by artificial intelligence (AI) technologies. These innovations have revolutionized traditional practices by enabling sophisticated monitoring and analysis techniques. We present an application of AI in beekeeping, where a remote monitoring system equipped with sensors and algorithms provides unprecedented insights into hive conditions and bee performance. Bees are essential to the global surroundings and, more critically, to the survival of many crops because of their capacity to pollinate flowers. Due to a disease known as Colony Collapse Disorder, beekeepers have recently witnessed significant losses to their controlled honeybee colonies as their colonies are declining. Most bees mysteriously leave the hive when a colony gets CCD. A beekeeper has to fully comprehend the state and actions of a hive to take proper care of bees. To fulfil these requirements, we developed an Artificial Intelligence based beehive that is affordable, entirely scalable, and straightforward to use. The temperature, humidity, weight and noise levels inside the hive are all measured by the equipment. An Infrared (IR) sensor is mounted on the exterior of the hive to count the number of bees and monitor activity at the entrance. The data from the sensors and IR uses to indicate the condition of the hive by monitoring usual activities of bees. The sensors in this system continually collect data, which it then sends to a remote server for analysis. All data is analysed by algorithms, which then alert the beekeeper to any changes in the health of the hives. This contribution improves user-experience and amateur for keeping honeybee and gives way for interoperability between the available simulation (humidity, weather, bee colony behaviour etc). Through rigorous experimentation and analysis, we achieved remarkable accuracies of 97.49%, 98.49%, and 98.37% for Multi-layer Perceptron, Regression, and Random Forest, respectively.