DeepReef
Coral Bleaching Detection

DeepReef which is a Multimodal deep learning system for coral bleaching detection combining underwater digital image processing, a fine-tuned NOAA marine science classifier, XGBoost environmental modeling into a late fusion mode for predictions interpreted by gorq LLM.

It combines three independent models into a single diagnostic pipeline: a custom underwater image enhancement module that corrects blue-shift, removes speckle noise, and boosts local contrast before any model sees the image;
a fine-tuned NOAA National Marine Fisheries Service coral bleaching classifier trained on 922 coral photographs; and an XGBoost environmental model trained on reef site observations with physics-informed feature engineering based on NOAA bleaching alert thresholds.
Predictions from both models are fused using confidence-weighted combination and explained in plain language via a Groq LLM.
Simple and intuitive interface for fast experimentation