Skip to main content
BASELINE.ENVEnvironmental Monitoring
AI Detection

Species Identification. Done Automatically.

Upload camera trap images and field audio recordings. Baseline AI identifies species and pre-populates your review queue — so ecologists confirm results rather than start from scratch.

Pipeline

From Upload to Verified Record in Six Steps

Every image and audio file moves through a consistent pipeline — automated where possible, expert-reviewed where it matters.

01

Upload

Drag camera trap batches and acoustic recordings into the media library — from a field laptop, USB drive, or shared folder. No special preparation needed.

02

Detect

MegaDetector scans every camera trap image and separates animal detections from empty frames, people, and vehicles. Runs automatically on upload.

03

Identify

SpeciesNet classifies detected animals by species. Perch v2 processes acoustic recordings in parallel. Camera and acoustic jobs run overnight in bulk.

04

Queue

Every detection enters a structured review queue with the AI suggestion, confidence score, and bounding box pre-populated. Common species are grouped for batch approval.

05

Review

Ecologists confirm, correct, or reject detections using keyboard shortcuts. Hundreds of images reviewed per hour instead of per day.

06

Record

Confirmed detections become verified species observation records linked to your project species list — ready for regulatory export and reporting.

0%

Less time on camera trap review

AI pre-sorts images so ecologists only review detections, not empty frames

$0

Per-image processing cost

All AI models run on your infrastructure — no cloud fees, no limits

0%

Data stays on your servers

AI processing runs on your own infrastructure — survey media never leaves your network

Detection Models

AI Models Built for Australian Field Surveys

Camera trap and acoustic survey types are covered — every model included in your subscription, running on your own infrastructure.

MDAnimal Detection

MegaDetector

Scans every camera trap image and instantly separates images with animals from empty frames — so your team never wastes time clicking through blank photos. Built on the Apache-licensed MegaDetector v6 stack, self-hosted on your infrastructure with no per-image costs.

  • Sorts thousands of images overnight while the team sleeps
  • Highlights exactly where the animal appears in the frame
  • Filters out people, vehicles, and triggered-but-empty images
  • Your images stay on your servers — no cloud upload required
SNSpecies Classification

SpeciesNet

Identifies the wildlife species in each detected image and suggests a match — giving your ecologists a verified starting point rather than a blank record. Built by Google, recognises 2,000+ species with location-aware geofencing to narrow suggestions to your survey area.

  • Recognises 2,000+ wildlife species from camera images
  • Location-aware geofencing narrows suggestions to your area
  • Threatened and protected species automatically flagged
  • Runs automatically after animal detection — no extra steps
PV2Acoustic Species Identification

Perch v2

Processes hours of acoustic recordings overnight and identifies bird and wildlife calls — so your ecologists review a pre-classified queue each morning. Built by Google and covering 10,000+ species.

  • Processes entire recording libraries without manual intervention
  • Identifies acoustic species candidates from field recordings automatically
  • Batch-confirm high-confidence identifications in a single click
  • Uncertain calls flagged separately for expert review
Species Records

Detections Always Link to Your Species List

AI models use their own naming conventions. Baseline automatically bridges the gap — matching detections to your project species list through a cascading match system, so your records stay clean without manual data correction.

1

Confirmed Mappings

Reviewer-verified matches that improve automatically over time for your project

L1
2

Exact Name Match

Direct scientific name match to your project species list — the most common case

L2
3

Taxonomic Synonym

Handles reclassified species names so old and new names both resolve correctly

L3
4

Common Name Alias

Matches regional and colloquial species names used by different AI models

L4
5

Genus-Level Match

Links detections at genus level when species identification is uncertain

L5
6

Family-Level Fallback

Broadest fallback for rare or unusual detections that need expert review

L6

See the Pipeline in Action

All AI models run on your own infrastructure. Survey images and audio are never sent to the cloud, there are no per-image charges, and you can process as many surveys as you need.

0

AI Models Included

0+

Camera Trap Species

$0

Per-Image Cost

0%

Data Sovereignty