LRMR: Implementations of Structured Matrix Representation Methods

=========================================================================== *** The code is run on MATLAB R2012a-2014a on Windows 7 or Linux - 32/64 bit.

Usage:

1. put the input datasets into file '\INPUT_IMG' 将数据集文件夹放入 '\INPUT_IMG' 中
2. put the groundtruth images into file '\GROUND_TRUTH' under the folder that same name as dataset 将数据集GROUND TRUTH文件放入 '\INPUT_IMG' 中与数据集同名文件夹中
3. run 'compile.m' 运行 'compile.m'
4. adjust 'demo.m' add dataset to list datas 调整'demo.m'文件，将数据集名称添加入datas列表中
5. run 'demo.m' 运行 'demo.m'
6. the output saliency maps are stored in the file '\SAL_MAP' under the folder that same name as dataset 显著图将保存在'\SAL_MAP'中与数据集同名文件夹中
7. the evaluation results are stored in the file '\results' 评价结果将保存在'\results'中

Support Models

1. SMD: Structured Matrix Decomposition
2. BSMD: Bi-factorization method for SMD
3. TSMD: Tri-factorization method for SMD
4. BSMN: Bilinear Factor Matrix Norm minimization for SMD
5. DNN-SMD: Double Nuclear Norm for SMD
6. FNN-SMD: Frobenius/Nuclear Norm for SMD

Cition

@software{zhou2018lrmr,
  title  = {LRMR: Implementations of Structured Matrix Representation Methods},
  author = {Zhou, Houqan},
  url    = {https://github.com/Jacob-Zhou/LRMR},
  month  = sep,
  year   = {2018}
}