This Caffe fork was created on April 18, 2016. The changes to the original are listed below.
The loss function computed in EuclideanLossLayer is changed to weighted Euclidean loss, which enables coordinate specific loss calculation.
Changed the ImageDataLayer to support multiple labels so that it can be used for regression tasks. The input file supports the following format, where the values can be, for example, coordinates (x1, y1, x2, y2, ....)
01594.jpg 0.3284 0.8941 0.5021 0.7479 0.4534 0.5106
01741.jpg 0.7152 0.9104 0.6800 0.7024 0.6288 0.5456
01320.jpg 0.1612 0.4908 0.2821 0.5348 0.4835 0.3700
Added PosenetAccuracyLayer for logging accuracies and errors of coordinate predictions (xy) while training.
Integrated Faster R-CNN code so that this fork supports Faster R-CNN for object detection.
- After building Caffe, go to folder
matlab/posenet
with MATLAB and runsetup_posenet.m
to build Faster R-CNN and download trained models. - Run
demo_pretrain.m
to see how pretrained model performs on random evaluation images from MPII Human Pose dataset.
Caffe is a deep learning framework made with expression, speed, and modularity in mind. It is developed by the Berkeley Vision and Learning Center (BVLC) and community contributors.
Check out the project site for all the details like
- DIY Deep Learning for Vision with Caffe
- Tutorial Documentation
- BVLC reference models and the community model zoo
- Installation instructions
and step-by-step examples.
Please join the caffe-users group or gitter chat to ask questions and talk about methods and models. Framework development discussions and thorough bug reports are collected on Issues.
Happy brewing!
Caffe is released under the BSD 2-Clause license. The BVLC reference models are released for unrestricted use.
Please cite Caffe in your publications if it helps your research:
@article{jia2014caffe,
Author = {Jia, Yangqing and Shelhamer, Evan and Donahue, Jeff and Karayev, Sergey and Long, Jonathan and Girshick, Ross and Guadarrama, Sergio and Darrell, Trevor},
Journal = {arXiv preprint arXiv:1408.5093},
Title = {Caffe: Convolutional Architecture for Fast Feature Embedding},
Year = {2014}
}