Classification of Motion Regions with Convolutional Networks, Support Vector Machines, and Random Forests in Video-Based Analysis of Bee Traffic

Abstract

Bee traffic is the number of bees moving in a given area in front of a specific hive over a given period of time. Video-based bee traffic analysis has the potential to automate the assessment of bee traffic levels, which, in turn, may lead to the automation of honeybee colony health assessment. In this paper, we evaluate several convolutional networks to classify regions of detected motion as BEE or NO-BEE in videos captured by BeePi, an electronic beehive monitoring system. We compare the performance of several convolutional neural networks with the performance of support vector machines and random forests on the same image dataset