Sport Objects recognition

Help
2012-10-11
2012-12-24
  • xtremeredneck

    xtremeredneck - 2012-10-11

    I am working on a project module in which i need to recognize sport objects from the given image for this i am using Neuroph framework's image recognition module and follow this example (http://neuroph.sourceforge.net/image_recognition.html).So for this i first train the system with different basketball images with learning rate=0.5,momentum=0.7, and max error rate=0.01 and also use 12 hidden neurons then give a test image to recognize the basketball from the image But the problem is that it doesnt give me correct results So i am looking for some ideas/suggestions/help on how can i improve/correct my results

     
  • Zoran Sevarac

    Zoran Sevarac - 2012-10-11

    Hi,

    Try using learning rate 0.2
    Have you managed to train the network at all?

    Zoran

     
  • xtremeredneck

    xtremeredneck - 2012-10-12

    hi,
    Thanks for the prompt reply, i have used learning rate of 0.2 and its showing me Total Net Error of 0.0495 with 5400 iterations.Also at some point  i stop training the network because the Total Net Error isn't affected for certain amount of iteration

     

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:





No, thanks