In this paper, we propose to evaluate the use of pre-trained convolutional neural networks (CNNs) as a features extractor to perform classification of the genus Oliva Bruguière, 1789. These mollusks include species whose overall smooth and ellipsoidal shape encompasses a limited degree of variability. The hallmark of this natural group, a combination of low interspecific variability and high intraspecific variability, has been for more than two centuries a source of taxonomic instability, including but not limited to the status of subgenera, or alleged genera, included in the Oliva clade. In order to assist the Linnean systematics of controversial groups, whose taxonomy is largely based on morphological landmarks, the present study explores the potential effectiveness of deep learning models as feature extractors for classifying the genus Oliva. In particular, by leveraging a dataset of about 1,000 digital images of Oliva shells categorized into 9 distinct subgenres, we employed pre-trained convolutional neural networks (CNNs) for feature extraction. Subsequently, applying statistical learning algorithms on a training set, we conducted a statistical classification analysis to assess the potential of an automated classification procedure in accurately predicting the labels of a test set. Preliminary results suggest that employing a pre-trained VGG16 architecture for feature extraction yields an accuracy of approximately 82% in classifying the test images.