A Deep Learning based CNN framework approach for Plankton Classification

Planktons form the base of the Marine food chain. Phytoplanktons account for nearly 50% of all photosynthesis on earth and around 50% to 85% of the oxygen in Earth's atmosphere. The study of Zooplanktons and their habitat helps us in identifying the diet and migration patterns of various marine species. Since the impact of Planktons on life on Earth is so colossal, their study and identification have become a vital part of Marine Biology. Our research proposes a generic framework for the classification of various types of Planktons found across the world's oceans. We worked on classifying images into the following five categories- Diatoms, Crustaceans, Cyanobacteria, Cnidarians, and Molluscs. We have proposed a novel Framework architecture of Inception v3 as our Feature Extraction Model coupled with Convolution Neural Network as the classifier. Our model gives us the best accuracy of 99.5%.