mcp · Global

A.Russo-Marine-AI

Marine Biodiversity AI - Capstone Project

Claim this listing

Represent A.Russo-Marine-AI? Send a claim for review so we can discuss listing updates. Claims do not automatically grant ownership or verification.

Data offered docs-verified: marine species classification from image (196 species) · OBIS occurrence time series with anomalous years flagged · OBIS occurrence geographic map of sighting points · taxonomic and scientific-caption knowledge base (RAG)
Coverage llm-inferred: Global
Formats docs-verified: MCP stdio
Access docs-verified: No authentication (draft MCP server). Local install from the repository (pip requirements, ChromaDB, LM Studio local LLM). Licence not stated.

Use cases llm-inferred: Screening species IDs from field or survey photos against the 196-class classifier, with top-3 confidence for QA review · Baseline OBIS occurrence trend check for a marine species in a coastal EIA desk study · Mapping OBIS sighting points as a marine receptor screening layer
Audience llm-inferred: marine ecologists · environmental consultants · GIS analysts · ML engineers · marine biologists

Constraints docs-verified: Draft MCP server: no authentication, no production hardening, stdio transport only · Copernicus Marine integration is documented but not implemented and requires separate registration · Classifier validation accuracy is 56.1% · Only 145 of 196 species have full environmental coverage · Seahorse and pipefish are excluded from the environmental module due to a taxonomy error in ORCA metadata · Anomaly detection does not compensate for OBIS platform growth over time · Generic labels such as fish and coral have notably lower F1 · Requires running notebooks 01-06 to download data and train the model, plus LM Studio and ChromaDB running · No UK-specific coverage is stated in DOCS

Auto-extracted, unverified fields: tools, install, auth. Check upstream documentation.

Daily discovery 2026-10-16: GitHub search; verified via GitHub API. Tools/install/auth auto-extracted from README.

Metadata observed 14/10/2026 · daily-2026-10-16 (github-search)