• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 99 subscribers
  • Views 610 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA2: TIDL: Using Caffe_Jacinto framework

Srinidhi S46
Prodigy 160 points

Part Number: TDA2

I want to get the Caffe-jacinto framework and run training on out custom dataset. So train around for 100 images, Will we be needing a GPU or a simple non-GPU system will work. 

  • 0
    TI__Genius 11546 points
    You can get the caffe-jacinto framework and scrits by clonning the following two repositories:
    github.com/.../caffe-jacinto
    github.com/.../caffe-jacinto-models

    Please navigate to those links in your browser and read through the documentation provided to understand the usage. If you face any specific issue related to the code or scripts, you can file an issue using the github repository "issues" link in the above repositories.
  • 0 in reply to Manu Mathew
    Prodigy 160 points

    Yes, I understand that. I wanted know if a system without a GPU is used, approximately how long it'll take to train for say 100 images? Is it recommended.

  • 0 in reply to Srinidhi S46
    TI__Genius 11546 points
    Training on CPU can be 30x or more slower compared to training on GPU. Exactly how much time will be needed will depend on the size of the image, complexity of the network and the problem that you are trying to solve. Why don't you try to train on CPU - that will give you an estimate of the time required.

    But whether you conclude to use CPU or GPU for training, I would highly recommend to more images for training. Deep learning cannot learn to generalize with just 100 images and any model or conclusion that you make using that dataset may not work outside the laboratory.