DetectionEvaluationJ: A Tool to Evaluate Object Detection Algorithms

Abstract

Object detection is an area of computer vision with applications in several contexts such as biomedicine and security; and it is currently growing thanks to the availability of datasets of images, and the use of deep learning techniques. In order to apply object detection algorithms is instrumental to know the quality of the regions detected by them; however, such an evaluation is usually performed using ad-hoc tools for each concrete problem; and, up to the best of our knowledge, it does not exist a simple and generic tool to conduct this task. In this paper, we present DetectionEvaluationJ an open-source tool that has been designed to evaluate the goodness of object detection algorithms in any context and using several metrics. This tool is independent from the programming language employed to implement the detection algorithms and also from the concrete problem where such algorithms are applied.