Open Access

Fig. 6

image

Download original image

Images showing the effect of negative training on an R-CNN. In this comparison we are considering only the sizes and positions of the bounding boxes, as the R-CNN has classified them all as lenses with confidences ≥90%. The top image has had no negative training, while the bottom image has. Negative training has had a significant positive effect as all false positives have been removed, and the correct label box has been cropped significantly. Credit to ESA for the original image.

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.