Skip to content

xiaodiu2010/templete-tensorflow

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MMVC LAB Project Templete

This project is a training templete based on TensorFlow and Slim. It is designed for rapid model development and iteration by providing flexible infrence and all popular based network and pretrained model. Currently, maintained by: Daitao Xing ([email protected]), Mengwei Ren([email protected]) and Liang Niu([email protected]) from MMVC Lab, NYU/NYU Abudhabi.

Features

  • Data Loader
    Focus on performance and flexiability of data pipeline. User can define their own pipeline using native python code without concerning the details of feeding data.
    The way of loading data always becomes the bottleneck of GPU usages. The speed of loading data in this templete is much efficient than feed_dict way or queue way. It runs 1.2~5x faster than the equivalent code because the threads will prefetch the batches and trainers don't need to wait for queues.
  • Based Network and pretrained model
    The templete has included all based network used for classification task such as vgg, inception, resnet and so on. All those codes are from slim project so you can easily restore the pretrained model provided on slim website.
  • Multi-GPU Support training models using multi-gpus without wasting speed.
  • Restore checkpoints and ignore some variables
  • Flexible
    User can define their summaries, evauation and callback hooks.
  • Evaluation
  • Models for sgementation task Only contains u_net, more is coming soon~

This templete is under developing

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages