Databases
StackOverflow2013
Postsdbo
PO
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PO
    primarykey
    data
    text
    <p>If you insist on a neural network...</p> <p><strong>Simplify your prerecorded input</strong></p> <ul> <li>For each 3d point list create 3 black and white 2d images of your points (drawing a line between each adjacent point)</li> <li>Top-down (x,z)</li> <li>Forward-back (x,y)</li> <li>Left-right (z,y)</li> </ul> <p><strong>Simplify your user input</strong></p> <ul> <li>Create the same 3 black and white images (explained above)</li> <li>Find the bounding box of the points and scale the image to a uniform size</li> <li><ul> <li>Having your user input, scaled to same size as your prerecorded input will facilitate recognition</li> </ul></li> </ul> <p><strong>Create neural networks for image recognition</strong></p> <ul> <li>Create a trained network for your prerecorded (3 images for each)</li> <li>This seems like it would work <a href="http://neuroph.sourceforge.net/image_recognition.html" rel="nofollow">http://neuroph.sourceforge.net/image_recognition.html</a></li> <li>Build your neural nets from the black and white images (that represent the 2d mapping of your 3d points onto the 3 perpendicular planes)</li> </ul> <p><strong>Experiment with your matching bounds/constraints</strong></p> <ul> <li>IF Top2Bottom >= 0.75 AND Forward2Back >= 0.75 AND Left2Right >= 0.75 </li> <li>THEN IsAMatch = True</li> <li>ELSE IsAMatch = False</li> </ul> <hr> <p>I would probably just generate a large 3d corridor (think wide 3d tube, or a collection of connected cylinders) that represented the extremes of my prerecorded points then scale the user input points then check if the user input points fall within the matching corridor. </p> <hr> <p><strong>EDIT</strong> Very odd that the question itself was down-voted and closed, then days later my answer was down-voted. Realistically, the question was likely closed due to a failure of group think on the part of the moderators. Maybe it then looked a little awkward to have a closed question with a +1 "Accepted" answer, so this answer was subsequently down-voted (twice). </p> <p>From my experience, the only way to ask a question like this is to either slap a large bounty on it, or re-word it in a way that won't offend the fragile sensibilities of the moderators (most of the time they do a good job and don't overstep...).</p>
    singulars
    1. This table or related slice is empty.
    1. PONeural networks for a 3d space value linked list matching in java
      singulars
      1. PTQuestion
    1. PTAnswer
    1. USLouis Ricci
    1. USLouis Ricci
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PONeural networks for a 3d space value linked list matching in java
      singulars
      1. PTQuestion
    1. This table or related slice is empty.
    1. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTAcceptedByOriginator
    2. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTDownMod
    3. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    1. COfor the explanation with 3 images template matching algorithm would @LastCoder the idea with 3 images looks great will implement it as well, however I am thinking of using the Encog not the neuroph as it allows image analysis on the fly. Could you please tell me more about the 3D corridor idea ? Akonkagva
      singulars
      1. PO
      1. USPeter
    2. CO@user1165088 - consider a 3d tunnel/tube, it would be a collection of (Vector3, Radius) pairs. Now you have a user generated list of 3d points you stretch/scale those points to fit the bounds of your 3d tunnel then you see if those points lie within the tunnel or not using the distance formula.
      singulars
      1. PO
      1. USLouis Ricci
    3. COthank you for great ideas, I will implement them both !
      singulars
      1. PO
      1. USPeter
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload