Stream automatic detection with convolutional neural networks

Open Access

Table 4

SAD-CNN application on inclined disc.

SAD	Precision	Recall	F1-score	Support
Not Det	0.88	0.29	0.44	624
V-1 Det	0.60	0.96	0.74	692
Average	0.74	0.63	0.59	1316
Accuracy			0.65	1316

Not Det	0.88	0.58	0.70	624
V-2 Det	0.71	0.93	0.80	692
Average	0.79	0.75	0.75	1316
Accuracy			0.76	1316

Not Det	0.84	0.81	0.82	624
V-3 Det	0.83	0.86	0.84	692
Average	0.83	0.83	0.83	1316
Accuracy			0.83	1316

Notes. Summary of the values in the metrics obtained for the dataset with a total of 1316 random models at a random inclination, rotation, and snapshot time. The values obtained for each version of the convolutional neural network respectively.

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.