Ensemble Machine Learning-Based Egg Parasitism Identification for Endangered Bird Conservation.

The phenomenon of intraspecific nest parasitism is of a great interest to biologists because it helps in the conservation of critically endangered birds like the Slender-billed Gull. Indeed, upon detecting a parasitic egg in the nest, the Slender-Billed Gull female abandon it. This behavior results in loosing large number of future birds and leads to its extension. So, a nest cleaning from parasitic eggs has been an urgent need. Therefore, in this paper, we suggest a Slender-Billed parasitic egg identification system from the egg visual information to clean the nest and avoid the female leaving. Encouraged by the success of the machine learning models in pattern recognition and classification, we have built a system which extracts the most important egg visual features and classify them using a set of machine learning models which have been aggregated together to attain a high parasitic egg identification accuracy. In fact, a set of egg visual features have been extracted from the eggshell using the SpotEgg tool which have provided information about the egg coloration and patterning such as the egg shape, size, color and spottiness. Our model has been evaluated on a 31-nest dataset and has given an accuracy of 88.3% which has outperformed other machine learning models.