Regresar

Coccinellidae Beetle Specimen Detection Using Convolutional Neural Networks

Abstract:

In this work, we propose a ladybird beetle detector based on a deep learning classifier and the weighted Hausdorff distance as a loss function. The detector was trained and validated using ten-fold cross-validation method on a database composed of 2,633 wildlife images with ladybird beetles. Despite the detector performance was assessed using four metrics, the higher detection result of 98.25% was obtained using the precision metric. This result highlighted the successful performance of the implemented detector, and also, its competence for detecting ladybird beetles in different environments.