Prototype and machine learning integration for varroa mite detection in bees.
Abstract
In recent years, there has been a worldwide decline in the population of bees. Losses raise a serious concern because bees have an indispensable role in the food supply of humankind with Varroa mite infestation posing a significant threat to honeybee colonies worldwide including in Uganda. Early detection of these mites is crucial for effective management and prevention of colony losses. This project aimed to develop an embedded camera module supported by a machine learning algorithm for the process of early detection of Varroa infestations. This was achieved by using a machine learning algorithm based on Convolutional Neural Networks that tries to identify bees inside the beehives carrying the mite in real-time. To check the feasibility of the project a prototype was designed consisting of of two Esp32 cameras with one camera to provide a video stream or capture images for the different resolutions and the other to perform the varroa mite detection inside the beehive and the GSM Module sends a text message to the farmer’s phone once varroa mites are detected. The project was then implemented and tested and the results obtained demonstrate the effectiveness of the developed machine learning model in accurately detecting Varroa mites and the recommendations for the further improvement of the system are discussed and presented.