Self-hosted review UI for SpeciesNet wildlife classifier predictions
-
Updated
Nov 8, 2026 - Python
Self-hosted review UI for SpeciesNet wildlife classifier predictions
Fork of google/cameratrapai (SpeciesNet) with multi-crop per-detection classification, object filtering and MC-dropout uncertainty estimation — proposed in PR #78
AI assisted CCTV wildlife detection
Local HTTP service that classifies camera-trap photos and clips with Google SpeciesNet. Optional local model for the Trailwatch Home Assistant integration.
Small working project for running wildlife detection and classification on folders of camera-trap images and videos.
camRa: an R package for camera trap data management
Desktop app: process, review and publish wildlife recordings
Promotion gate, per-site trust report and label-free drift study for zero-shot camera-trap classifiers (BioCLIP 2, SpeciesNet). Cheap adapted challengers are scored site by site on the Orinoquía dataset with a frozen hold-out and paired bootstrap.
Automated wildlife detection pipeline for NFS/SMB-accessible storage. MegaDetector + SpeciesNet + web dashboard.
MCP server + always-on wildlife monitor for home IP cameras: a lightweight motion frontend gates MegaDetector + SpeciesNet to save compute, and visits go to a SQLite log Claude can query, with a web dashboard for review and tagging.
To associate your repository with the speciesnet topic, visit your repo's landing page and select "manage topics."