The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here
Driving myself crazy trying to create neural net!
Contributor II
I posted two other questions related to this issue, and have gotten no responses. I promise this will be my last on this issue but I would really like to get some suggestions and I KNOW there is someone on this board who can help.
I am trying to create a neural net that will recognize whether vehicles or people appear in images. I've used the ImageMiner-1.4.1 extension "global statistics" feature extractor and have gotten poor performance. I found another process called "thresholding" that should enable my neural network to easily distinguish the two classes of image. When I set a very low threshold, my images with no people or vehicles is very "washed out." When a vehicle or person is in the picture it is easily identifiable. All I want to do is submit these "thresholded images" to a neural network (labelled true or false by the different folders they are stored in) to train it to recognize the difference. But I can't seem to figure it out.
I've tried loading my images as color images and converting to grayscale (which is all I need and which is the format required by the thresholding operator), but my neural net requires an "example set" and the thresholding operator does not produce an example set, and I cannot find an operator that will take the thresholding output and convert it, transform it, or otherwise metamorphose it into an example set. Likewise, I've tried loading the images as grayscale to begin with, but I have the same problem of having the incorrect inputs or outputs.
Any help is appreciated.
I am trying to create a neural net that will recognize whether vehicles or people appear in images. I've used the ImageMiner-1.4.1 extension "global statistics" feature extractor and have gotten poor performance. I found another process called "thresholding" that should enable my neural network to easily distinguish the two classes of image. When I set a very low threshold, my images with no people or vehicles is very "washed out." When a vehicle or person is in the picture it is easily identifiable. All I want to do is submit these "thresholded images" to a neural network (labelled true or false by the different folders they are stored in) to train it to recognize the difference. But I can't seem to figure it out.
I've tried loading my images as color images and converting to grayscale (which is all I need and which is the format required by the thresholding operator), but my neural net requires an "example set" and the thresholding operator does not produce an example set, and I cannot find an operator that will take the thresholding output and convert it, transform it, or otherwise metamorphose it into an example set. Likewise, I've tried loading the images as grayscale to begin with, but I have the same problem of having the incorrect inputs or outputs.
Any help is appreciated.
Tagged:
0
Answers
the image mining extension has not been developed by Rapid-I, and I personally have never used it. Of course this forum nevertheless is a good starting point for help, but as your posts did not yield any answers, you may want to contact the extension's authors directly. There is a link to their website on the extension's page in the market place.
Good luck!
-Marius