Skip to content
/ CNN Public

CNN for image recognition based on CIFAR-10 dataset using tensorflow and serializing dataset in tfrecords datatype

Notifications You must be signed in to change notification settings

tetapunto/CNN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 

Repository files navigation

Dependencies

  • Python 3.5 or +
  • tensorflow 1.8
  • tensorboard 1.8

How to use

Download and create input file in .tfrecords

If you have already downloaded the *tar.gz file create a directory Input inside the directory CNNModel and put the .tar file inside it.
Then call the script file_creator, it will extract and transform the files.

Otherwise calling the script should automatically download the file, but this operation could fail for certain OS.

    $ python file_creator.py

Train Model

Move inside CNNModel directory and run

    $ python train.py

If a model already exists and you want to train from there:

    $ python train.py --path path_to_model/model_name.ckpt

If you want to restart from the last checkpoint

 $ python train.py --last True

Evaluate Model on Test Samples

Move inside CNNModel directory and run

    $ python model_accuracy_test.py

Visualize training on tensorboard

Move inside CNNModel directory and run

    $ tensorboard --logdir ./logs

About

CNN for image recognition based on CIFAR-10 dataset using tensorflow and serializing dataset in tfrecords datatype

Topics

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages