Detection and Identification Rate at rank K (DIR)

Rank

The rank of a certain sample or templatate is its position in the list of the matched templates in the gallery.

Detection and Identification Rate at rank K (DIR)

We are in the case of identification open set.

$D I R$ at rank $k$ measures the probability of correct identification at rank $k$ , meaning that the correct subject is returned at position $k$ in the list of the matched templates in the gallery.

Formally:

D I R (t, k) = \frac{∣ { P _{j} : r ank ( p _{j} ) \leq k , s _{ij} \geq t , i d ( g _{i} ) = i d ( p _{j} )} ∣\forall p _{j} \in P _{G}}{∣ P _{G} ∣}

Note that $FRR (t) = 1 - D I R (t, 1)$ .

👨🏽‍💻 Domiziano's Notes

Explorer

Detection and Identification Rate at rank K (DIR)

Rank

Detection and Identification Rate at rank K (DIR)

Graph View

Table of Contents

Backlinks