Transfer Learning for Arthropodous Identification and its Use in the Transmitted Disease Diagnostic.

Outdoors’s activities and sporadic nature getaways are becoming more and more common in recent years. Warm and humid climates without extreme temperatures favor insects or small organisms to live (and proliferate), which can cause potentially serious health problems if we do not have a minimum knowledge of what to do if we are bitten or stung. One of such concerning animals are the arthropodous. The objective of this work is to provide doctors and patients a machine learning-based tool to obtain a fast initial diagnostic based on a picture of the specimen which bit them. The developed model achieved over a 93% accuracy score based on a dataset of 493 color images. Three species have been categorized and analyzed, and the possible diseases they may transmit identified. The proposed system is effective and useful for a future real-life integration into a platform.