Using MobileNetV2 Deep Convolutional Neural Networks and Transfer Learning for Shell and Pebble Classification

In order to classify shells and pebbles, this study investigates the use of MobileNetV2 deep convolutional neural networks (CNNs) and transfer learning approaches. We train the network using a dataset of shell and pebble pictures, then use the pre-trained MobileNetV2 model as a feature extractor to refine the network and create a customised classification model. Future developments in the geological, coastal and environmental sciences might benefit from the use of MobileNetV2, deep convolutional neural networks (CNNs), and transfer learning in the categorization of shells and pebbles. Through transfer learning, the model can more effectively identify and access between various shell and pebble categories by utilising the information it acquired during its training on a wide range of pictures. Extensive tests are conducted to assess the efficacy of the suggested technique, revealing encouraging outcomes in terms of categorization efficiency and accuracy. With 87% accuracy, the proposed MobileNetV2 model predicts and classifies shells and pebbles with precision. By demonstrating the flexibility and effectiveness of MobileNetV2 in a particular area, the work advances the field of picture classification and opens the door to further developments in automated classification systems for environmental and geological applications. Further deep learning research might result in the creation of increasingly sophisticated model architectures designed especially for the categorization of geological and environmental objects. Accuracy and efficiency might be further enhanced by innovations in model design.