On Video Analysis of Omnidirectional Bee Traffic: Counting Bee Motions with Motion Detection and Image Classification

Abstract

Omnidirectional bee traffic is the number of bees moving in arbitrary directions in close proximity to the landing pad of a given hive over a given period of time. Video bee traffic analysis has the potential to automate the assessment of omnidirectional bee traffic levels, which, in turn, may lead to a complete or partial automation of honeybee colony health assessment. In this investigation, we proposed, implemented, and partially evaluated a two-tier method for counting bee motions to estimate levels of omnidirectional bee traffic in bee traffic videos. Our method couples motion detection with image classification so that motion detection acts as a class-agnostic object location method that generates a set of regions with possible objects and each such region is classified by a class-specific classifier such as a convolutional neural network or a support vector machine or an ensemble of classifiers such as a random forest. The method has been, and is being iteratively field tested in BeePi monitors, multi-sensor electronic beehive monitoring systems, installed on live Langstroth beehives in real apiaries. Deployment of a BeePi monitor on top of a beehive does not require any structural modification of the beehive’s woodenware, and is not disruptive to natural beehive cycles. To ensure the replicability of the reported findings and to provide a performance benchmark for interested research communities and citizen scientists, we have made public our curated and labeled image datasets of 167,261 honeybee images and our omnidirectional bee traffic videos used in this investigation.

Application of Digital Particle Image Velocimetry to Insect Motion: Measurement of Incoming, Outgoing, and Lateral Honeybee Traffic

“…The DPIV-based bee motion estimation algorithm presented in this article improves our previously proposed two-tier method of bee motion counting based on motion detection and motion region classification [6] in three respects: (1) it can be used to measure not only omnidirectional but also directional honeybee traffic; (2) it does not require extensive training of motion region classifiers (e.g., deep neural networks); and (3) it provides insect-independent motion measurement in that it does not require training insect-specific recognition models. Our evaluation results are based on four 30-s videos captured by deployed BeePi monitors.…”

“…The video data for this investigation were captured by BeePi monitors, multi-sensor EBM systems we designed and built in 2014 [4], and have been iteratively modifying [6,21] since then. Each BeePi monitor consists of a raspberry pi 3 model B v1.2 computer, a pi T-Cobbler, a breadboard, a waterproof DS18B20 temperature sensor, a pi v2 8-megapixel camera board, a v2.1 ChronoDot clock, and a Neewer 3.5 mm mini lapel microphone placed above the landing pad.…”

“…Thus, accurate measurement of honeybee traffic in the vicinity of the hive is fundamental for electronic beehive monitoring (EBM) systems that continuously monitor honeybee colonies to detect and report deviations from the norm. Our EBM research (e.g., [4][5][6]), during which we watched hundreds of field-captured honeybee traffic videos of three different bee races (Carniolan, Italian, and Buckfast), leads us to the conclusion that honeybee traffic in the vicinity of a Langstroth hive [7] can be divided into three distinct classes: incoming, outgoing, and lateral. Incoming traffic consists of bees entering the hive either through the landing pad or holes drilled in supers.…”

“…Lateral traffic consists of bees flying more or less in parallel to the front side of the hive (i.e., the side with the landing pad). We refer to these three types of traffic as directional traffic and juxtapose it with omnidirectional traffic [6] where bee motions are detected regardless of direction.…”

“…In this investigation, we contribute to the body of research on insect and animal flight by presenting a bee motion estimation algorithm based on digital particle image velocimetry (DPIV) [8] to measure levels of directional honeybee traffic in the vicinity of a Langstroth hive. The algorithm does not compute honeybee flight paths or trajectories, which, as our longitudinal field deployment convinced us [4,6], is feasible only at very low levels of honeybee traffic. Rather, the proposed method uses DPIV to compute motion vectors, analytically classifies them as incoming, outgoing, or lateral, and uses the classified vector counts to measure directional honeybee traffic.…”