Message:

<p>[QUOTE=&quot;messydesk, post: 3908005, member: 1765&quot;]Heritage, Stacks/Bowers, Great Collections.&nbsp; You should be able to find lots of 09-S VDB And 14-D in lower grades.&nbsp;&nbsp; Stay away from 22, because those will cause problems with your training.&nbsp; </p><p><br /></p><p>You will need to use photos from more than one source or you will have ground truth data that is corrupted by how photos from a single source tend to turn out.&nbsp; Another thing you should consider is using a high-pass filter on the images to remove or reduce the luster or sheen in the photos of the coins, as these are prominent features in photos that you don&#039;t want influencing the training.&nbsp; At the time of inferencing, you would apply the same filter to your test data.</p><p><br /></p><p><br /></p><p>The problem is that you have to select enough photos of each grade that the network can differentiate the grades.&nbsp; Luster and very small features can make a big difference in these grades.&nbsp; The source of your test data will be even more important here, since variations in photo technique can have a large impact on the images.&nbsp; I think you&#039;ll have better luck with circulated grades.</p><p><br /></p><p>Good luck, and let us know how this works out.[/QUOTE]</p><p><br /></p>

Your name or email address:

Do you already have an account?

• No, create an account now.
• Yes, my password is:
• Forgot your password?

Stay logged in