I have a data base of 3D models along with a set of shape measures (length, diameter, etc). Is there some supervised learning method (metric learning, deep learning, dictionary learning) which is able to generate the 3D model from the measured parameters?
The answer is a qualified yes depending on the information contained on the database and how it relates to your 3d model.
Could you provide more information on the database(either here or through direct communication)
Arturo Geigel Thank you for your attention to this matter. The application is to retrieve an 3D approximate model of body parts (e.g. a thumb) to design exoskeletons/prothesis/orthesis from antropometric measures (e.g. thumb diameter, length, angulation, etc).
Daniel Prado de Campos
The objective of training your desired platform is twofold:
I would keep both goals independent of each other and train two separate systems (That is my personal preference so that tracing where the optimization needs to take place is easier to identify)
The first part can be as simple as a Backpropagation NN with encoded parameters to output image index numbers and the second is an NN along the lines of :
http://3d-r2n2.stanford.edu/
I would need more information on specifics to provide more details on implementation
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning