AntCounter Software: Counting Leaf-Cutting Ants Was never so Precise, fast and Easy

Abstract

One of the main concerns when collecting data is the bias stemming from the researchers’ expertise and training. The precision and accuracy of the data being collected can vary depending on the focus, concentration, mood, and experience of a researcher, and this can generate errors in data collection that can influence the results. We present a software for counting leaf-cutting ants that tracks the number of ants on previously recorded videos. The software was developed in the Python language and uses an image processing library called OpenCV (Open Source Computer Vision). It is based on a low-resolution video (640 × 480 pixels) taken with a camera mounted on a tripod and orientated perpendicularly to the ground. This open access program uses the movement of ants to identify their position and track them. It counts the number of ants that pass through a given point in two directions with respect to the nest entrance: incoming and outgoing. Comparisons between AntCounter and conventional methods used to count ants show that the program has higher accuracy than traditional counting in real time while reducing data collection time. We also report the error in the data collected by AntCounter. Using AntCounter the researcher reduces both the time of data collection in the field and the time to process data. It is also possible to improve the sampling effort (e.g., increase the number of samples, the number of experimental units), and measure the ants for longer periods of time, and at lower prices.