ORION LIÇI, ANA KTONA
Abstract
For better biodiversity conservation we must accurately monitor the insect population. As we already know, insects play a key role as pollinators. This is not, however, the only useful and crucial process in which they are involved. Other useful processes include decomposition, pest control and of course, being a food source for other species (Sánchez-Bayo & Wyckhuys, 2019). It is very disturbing that insect populations are in decline globally. This decline can lead to many environmental consequences, which will subsequently translate into economic problems. Traditional methods of monitoring are labor-intensive, time-consuming, and prone to error. This paper explores the potential of automatic insect counting systems, reviewing the current state of technologies, including visual, optical, acoustic, and hybrid sensor-based methods. Special attention is given to AI-driven approaches such as YOLO (Redmon et al., 2016) and TensorFlow.js, (Smilkov et al., 2019) which enable real-time detection and classification in the field.
Key words: Biodiversity conservation, AI-Driven methods, YOLO (You Only Look Once); Real-time detection.