Plankton classification benchmark

I recently worked on a benchmarking framework for plankton image classification across multiple imaging systems (IFCB, FlowCam, ISIIS, ZooCam, ZooScan, UVP6). The project compares Convolutional Neural Networks (CNN) and Random Forest classifiers, highlighting the importance of feature extraction vs. classifier complexity.

The repo includes:

• Data preparation scripts for multiple plankton imaging datasets (SeaNoe, WHOI IFCB)

• Training pipelines for CNNs (EfficientNet, MobileNet) and Random Forests

• Tools to extract deep features and apply dimensionality reduction (PCA)

• Evaluation scripts for performance metrics and visualization

This work supported the paper “Benchmark of plankton images classification: emphasizing features extraction over classifier complexity”.

Check out the repo here: https://github.com/ThelmaPana/plankton_classif