Marine Jellyfish Species Classifier
Automated Species Identification using Convolutional Neural Networks
Key contributions
- Custom CNN architecture extracting morphological features from underwater imagery.
- Applied spatial data augmentation to counter water turbidity and illumination variance.
- Automates tedious marine taxonomic categorization for oceanographic research.
Applied Machine Learning Engineering Specification
Structured technical analysis conforming to production machine learning standards.
Problem Formulation
Oceanographic biologists face massive manual backlogs indexing underwater video and camera traps to monitor marine biodiversity and jellyfish blooms.
Data Ingestion & Feature Space
Multi-species underwater imagery dataset featuring distinct jellyfish classifications (Moon, Compass, Lion's Mane, Barrel) captured across diverse ocean conditions.
Preprocessing & Methodology
Implemented extensive image transformations including rotation, shear, horizontal flips, and zoom alongside channel normalization.
Model Architecture & Training
Designed multi-tier Convolutional Neural Network with progressive feature maps (32 → 64 → 128), Batch Normalization, Dropout (0.3), and Softmax classification.
Validation & Evaluation Protocol
Tracked categorical cross-entropy loss convergence, per-species confusion matrices, and precision/recall balance across classes.
Empirical Results & Discoveries
Achieved robust taxonomic separation across challenging morphometry, accurately identifying species despite motion blur and light refraction.
Production & Business Impact
Accelerates ecological research workflows by automating organism tagging in continuous underwater survey feeds.
Tech stack
Want to see the source code?
Access the complete Jupyter notebook, preprocessing routines, and evaluation logs.