Open Access

Table 1

Detected OVS for real-world recordings of one person. In total, 326 087 intervals with a duration of 125 ms were classified. The manual labels are denoted as “True”. OVS detected with the RF algorithm and with the algorithm described in [31] are denoted as “Estimated” (Est.).

OVD with RF algorithm
OVD described in [31]
Est. OVS Est. non-OVS Est. OVS Est. non-OVS
True OVS 19 660 (84.1%) 3717 (15.9%) 16 501 (70.6%) 6876 (29.4%)
True non-OVS 7645 (2.5%) 295 065 (97.5%) 10 738 (3.5%) 291 972 (96.5%)

Note: Bold: Percent correct for the comparison between manually labeled ground truth and OVS predictions are given along the diagonal of the confusion matrices.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.