Application of Image Processing for Gastropod Identification: Analysis of Shell Morphology

Gastropods, a diverse group of invertebrates with single shells, are often identified through their unique shell shapes and patterns. Computer-assisted taxonomic assessments from shell morphology can save time and eliminate human error in taxonomic studies on Gastropods. Recently computer-based applications have focused on species identification based on visual records of individuals, reducing errors and biases due to human interpretation. This study evaluated the applicability of data library created from images of 10 individuals of six different Gastropod species taken from four different angles using a stereo microscope for species identification. These images were processed into digital data with SqueezeNet and Inception v3 algorithms and analyzed using cosine distance and hierarchical clustering techniques. Analysis carried out with the library data showed that images of the same species in our library dataset were clustered together. When the analyses were repeated using visual records of the species being aimed for identification, it was observed that 90% of the newly considered individuals were classified under their respective species clades. This suggests that taxonomic identification is an applicable methodology that can be applicable quickly from images taken in field or laboratory conditions, potentially serving as a preliminary evaluation in advanced species identification studies.