First Intermediate Results

Two teams submitted their methods to the competition: IDIAP and NTNU.

Results of the PERIOCULAR TASK
IDIAP and NTNU submitted periocular methods.


Table 1. Results of the 5 methods submitted (error rate in %)
 MethodRank
 GF2
GFRR@GFAR=0.01
 EER
 IDIAP212.08
1.56
 NTNU1 33.16
2.58
 IDIAP12
3.712.38
 NTNU3 44.733.05
 NTNU2 54.80
3.05




Results of the IRIS TASK

NTNU submitted 3 iris methods.

Table 2. Results of the 3 methods submitted (error rate in %)
 Team
Rank 
GF2
GFRR@GFAR=0.01
EER
 NTNU3
 1
9.61
6.25
 NTNU1
 2
10.51
7.34
 NTNU2
 3
17.62
 11.38


1. An intermediate test set, with 20/40 users (for Periocular and Iris, respectively) was used to obtain the intermediate results.
2. All possible cross-sensor intra-class comparisons are implemented to evaluate the false non-match rate (FNMR).
3.
Three samples were selected from each class for each sensor to evaluate the false match rate (FMR).

4. The metric GF2 given by GFRR@ GFAR=0.01(
same as GFRR@ GFAR=1%) will be used as the performance indicator to rank the submitted algorithms. (EER will be used in case of a tie.)
5. 
The procedure to deal with failures is compliant to the international standardization documents (namely, ISO/IEC 19795-1:2006).
Therefore, the evaluation was made using the error metrics GFAR and GFRR that consider both the failure-to-enroll (FTE) and the failure-to-acquire (FTA) rates.

Generalized false reject rate and generalized false accept rate:

GFAR = FMR * (1 – FTA) * (1 – FTE)

GFRR = FTE + (1 – FTE) * FTA + (1 – FTE) * (1 – FTA) * FNMR

(where FTA is the Failure to acquire rate and FTE is the failure to enrol rate, both according the ISO document).







Comments