Institute Homepage
Institute Homepage DE Sign In

Back

Perceiving Systems Autonomous Motion Conference Paper 2016

Barrista - Caffe Well-Served

Thumb ticker sm christoph
Perceiving Systems
Christoph Lassner
Affiliated Researcher
Thumb ticker sm dkappler medium
Autonomous Motion
Daniel Kappler
  • Doctoral Researcher
Thumb ticker sm small image
Robust Machine Learning
Martin Kiefel
Thumb ticker sm petergehler copy
Perceiving Systems
Peter Vincent Gehler
  • Research Group Leader
Thumb

The caffe framework is one of the leading deep learning toolboxes in the machine learning and computer vision community. While it offers efficiency and configurability, it falls short of a full interface to Python. With increasingly involved procedures for training deep networks and reaching depths of hundreds of layers, creating configuration files and keeping them consistent becomes an error prone process. We introduce the barrista framework, offering full, pythonic control over caffe. It separates responsibilities and offers code to solve frequently occurring tasks for pre-processing, training and model inspection. It is compatible to all caffe versions since mid 2015 and can import and export .prototxt files. Examples are included, e.g., a deep residual network implemented in only 172 lines (for arbitrary depths), comparing to 2320 lines in the official implementation for the equivalent model.

Author(s): Christoph Lassner and Daniel Kappler and Martin Kiefel and Peter Gehler
Book Title: ACM Multimedia Open Source Software Competition
Year: 2016
Month: October
Project(s):
Bibtex Type: Conference Paper (inproceedings)
DOI: 10.1145/2964284.2973803
Event Name: ACM OSSC16
Event Place: Amsterdam
State: Published
URL: https://github.com/classner/barrista
Digital: True
Electronic Archiving: grant_archive
Attachments:

BibTex 

@inproceedings{barrista,
  title = {Barrista - Caffe Well-Served},
  booktitle = {ACM Multimedia Open Source Software Competition},
  abstract = {The caffe framework is one of the leading deep learning toolboxes in the machine learning and computer vision community. While it offers efficiency and configurability, it falls short of a full interface to Python. With increasingly involved procedures for training deep networks and reaching depths of hundreds of layers, creating configuration files and keeping them consistent becomes an error prone process.
  
  We introduce the barrista framework, offering full, pythonic control over caffe. It separates responsibilities and offers code to solve frequently occurring tasks for pre-processing, training and model inspection. It is compatible to all caffe versions since mid 2015 and can import and export .prototxt files.
  
  Examples are included, e.g., a deep residual network implemented in only 172 lines (for arbitrary depths), comparing to 2320 lines in the official implementation for the equivalent model.},
  month = oct,
  year = {2016},
  slug = {barrista},
  author = {Lassner, Christoph and Kappler, Daniel and Kiefel, Martin and Gehler, Peter},
  url = {https://github.com/classner/barrista},
  month_numeric = {10}
}