accessibility__skip_menu__jump_to_main

Full text: Automatic detection of boulders by neural networks

Boulder detection | 
59 objects at the test site, characterised by slopes 
ranging from 35° to less than 3.5°. However, most 
'dentihed boulders show slope values of over 4°. 
"he model running on the combined data set of 
Yackscatter, slope and depth detects 53 boulders 
Most of these boulders are also recognised in the 
sul manıntarı 
tawar 
slope data set. However, several potential boulders 
‘ound in the slope data set were not found by the 
combined model and vice versa, with examples 
shown in Fig. 7. Here, a comparison with the in- 
dependently recorded side-scan sonar data —- bar 
ing some uncertainty because of the positional 
9.6 
cu 
bl 
NW 
Mlgr 
{ 
Water column 
stratification 
*Nadir 
1 
Profile Overlap 
IS 9 I[ml-19 rt 
Profile Overlap 
= 79 
10.6 
Jadir 
1OW 
4 
Profile Overlap ! 
8} 
= RR 
Irfile Overlap 
1arlir 
Model confidence ® 
a 
0,2-0,4 
0.4-0.6 
® 06-08 
0 25 50m 
Fr 
X) 
Fig. 6: Boulders found by the models in the test area in the different data sets. For the SSS backscatter mosaic, 
magnified insets show the similarity of small boulders and artefacts due to water column stratification and near- 
verticral inridence, Refer to Fig, 2 for locatior 
AA 
119 — 06/2021
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.