Evaluation metrics

Evaluation Criteria

For the evaluation process, the output from your system should be an integer or a float from 0 to 100. This problem can be approached as a classification or regression problem, eitherway the results will be evaluated by fitting a normal distriution with the mean and standard deviation of the votes for each image:


Sample Image

Sample Label distribution

Each prediction is evaluated using the following formula,

Evaluation Formula

getting an error value between 0 (correct) and 1 (far from age). Not predicted images are evaluated with 1.

Evaluation Figure

Submission format

In this track, the participants should submit a ZIP file containing only a text file namedPredictions.csv (do not add any folder in the ZIP). Each line should contain a single identifier and the predicted age, separated by a semicolon, for example:



There are no news registered in Apparent age V1 (ICCV '15)