AI TrackerFrontier desk · AEST
Back to news
normalaitech

CounterFactual Retrieval for Evidence-Aware Individual Animal Identification

arXiv cs.AI · 5 Nov 2026, 04:19 AEST

Automated identification of individual animals is advancing with new evaluation methods that clarify the evidence behind matches. The Counterfactual Foreground Retrieval (CFR) protocol provides a comprehensive assessment of retrieval performance across different views.

Key facts

Read aloud

Automated identification of individual animals is advancing with new evaluation methods that clarify the evidence behind matches. The Counterfactual Foreground Retrieval CFR protocol provides a comprehensive assessment of retrieval performance across different views. Automated identification of individual animals is moving from small curated studies toward shared benchmarks and reusable foundation models. We address this with Counterfactual Foreground Retrieval CFR, an evidence-aware evaluation protocol. CFR evaluates the same query-gallery task under different views: full image, foreground-only, and shortcut-only. This turns a retrieval score into an evidence profile: how well the model performs on the original images, how much of that performance remains when only the animal is visible, and how much identity signal remains recoverable when the animal is removed. We release around zero.5M masks for thirty four datasets and code for mask generation and CFR evaluation.

Sources in this cluster