matconvnet-master
所属分类:其他
开发工具:Others
文件大小:2141KB
下载次数:0
上传日期:2018-11-25 20:46:49
上 传 者:
yeikingnan
说明: MatConvNet是一个实现用于计算机视觉应用的卷积神经网络(CNN)的MATLAB工具箱。它简单,高效,可以运行和学习最先进的CNN。包括若干示例CNN以对图像进行分类和编码。请访问主页了解更多信息。
(MatConvNet is a MATLAB toolbox implementing Convolutional Neural Networks (CNNs) for computer vision applications. It is simple, efficient, and can run and learn state-of-the-art CNNs.)
文件列表:
CONTRIBUTING.md (1562, 2018-07-29)
COPYING (741, 2018-07-29)
Makefile (12463, 2018-07-29)
doc (0, 2018-07-29)
doc\Makefile (4269, 2018-07-29)
doc\blocks.tex (32154, 2018-07-29)
doc\figures (0, 2018-07-29)
doc\figures\imnet.pdf (18884, 2018-07-29)
doc\figures\pepper.pdf (702358, 2018-07-29)
doc\figures\svg (0, 2018-07-29)
doc\figures\svg\conv.svg (68592, 2018-07-29)
doc\figures\svg\convt.svg (65347, 2018-07-29)
doc\figures\svg\matconvnet-blue.svg (6561, 2018-07-29)
doc\figures\svg\matconvnet-white.svg (6734, 2018-07-29)
doc\fundamentals.tex (33879, 2018-07-29)
doc\geometry.tex (16551, 2018-07-29)
doc\impl.tex (26897, 2018-07-29)
doc\intro.tex (17869, 2018-07-29)
doc\matconvnet-manual.tex (4412, 2018-07-29)
doc\matdoc.py (7192, 2018-07-29)
doc\matdocparser.py (11110, 2018-07-29)
doc\references.bib (4639, 2018-07-29)
doc\site (0, 2018-07-29)
doc\site\docs (0, 2018-07-29)
doc\site\docs\about.md (10722, 2018-07-29)
doc\site\docs\css (0, 2018-07-29)
doc\site\docs\css\fixes.css (2957, 2018-07-29)
doc\site\docs\developers.md (3393, 2018-07-29)
doc\site\docs\faq.md (7480, 2018-07-29)
doc\site\docs\figures (0, 2018-07-29)
doc\site\docs\figures\stn-perf.png (59846, 2018-07-29)
doc\site\docs\figures\stn-samples.png (32055, 2018-07-29)
doc\site\docs\functions.md (3336, 2018-07-29)
doc\site\docs\gpu.md (1148, 2018-07-29)
doc\site\docs\index.md (4440, 2018-07-29)
... ...
# Fast-RCNN demo
This folder contains an example implementation of Fast-RCNN [1] in
MatConvNet. The example trains and test on the PASCAL VOC 2007 data.
There are three entry-point scripts:
* `fast_rcnn_demo.m`: runs the original Caffe model imported in MatConvNet.
* `fast_rcnn_train.m`: trains a new model from scratch, using pre-computed proposals.
* `fast_rcnn_evaluate.m`: evaluates the trained model.
Note that the code does not ship with a proposal generation method, so
proposals must be precomputed (using e.g. edge boxes or selective
search windows).
The `fast_rcnn_demo.m` code should run out of the box, downloading the
model as needed.
To test the training code using the first GPU on your system, use
something like:
run matlab/vl_setupnn
addpath examples/fast_rcnn
fast_rcnn_train('train',struct('gpus',1)) ;
fast_rcnn_evaluate('gpu',1) ;
## References
1. *Fast R-CNN*, R. Girshick, International Conference on Computer
Vision (ICCV), 2015.
近期下载者:
相关文件:
收藏者: