文件列表:

COPYRIGHT (1474, 2013-05-25)
run_all.m (529, 2013-05-24)
script\classify.m (3463, 2013-05-24)
script\evaluation_acc.m (1077, 2013-04-18)
script\evaluation_VOC.m (1491, 2013-05-10)
script\run_classification.m (515, 2013-05-24)
script\voc\VOCevalcls.m (993, 2013-05-08)
script\voc\VOCinit.m (816, 2013-05-10)
** (1617, 2013-05-24)
** (619, 2013-05-24)
** (666, 2013-05-24)
** (1684, 2013-05-24)
** (1575, 2013-05-09)
** (1021, 2013-05-09)
** (980, 2013-05-09)
localdesc\calc_phow.m (1010, 2013-05-24)
localdesc\mykmeans.m (1776, 2013-05-10)
localdesc\visualword_phow.m (1176, 2013-05-24)
util\convert_15scene.m (1761, 2013-05-24)
util\convert_VOC.m (1391, 2013-05-24)
feat\dircode.m (604, 2013-05-09)
** (1364, 2013-05-09)
feat\sppart.m (1883, 2013-05-09)
coding\drc.m (572, 2013-05-09)
coding\myknn.m (509, 2013-05-24)

This is a MATLAB software package for image classification by histogram of oriented p.d.f gradients: T. Kobayashi, "BoF meets HOG: Feature Extraction based on Histograms of Oriented p.d.f Gradients for Image Classification," Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. -, 2013. Table of Contents ================= - Installation - Data - Usage - Reference Installation ============ Unzip hop***.zip in the folder you like. In a default setting, for SVM classification, this package requires SMO package which can be downloaded from http://staff.aist.go.jp/takumi.kobayashi/code/SMO.zip So, unzip SMO.zip and add the path into MATLAB cache. In addition, this also requires the vlfeat toolbox (http://www.vlfeat.org/) for extracting SIFT local descriptors. Data ==== You can apply these codes to any kinds of image datasets in which all image files are located in a directory; e.g., 0000.jpg, 0001.jpg. Information for class labels as well as cross-validation sets be descrived in 'cv_data_*.mat' which includes the following variables; classes - Class names ([1 x nclass], cell array) trainflist - File names excluding extension ('.jpg') for training, such as {'0000','0002',...} ([1 x ntrain], cell array) trainID - Class indicators by {-1,+1} for training ([ntrain x nclass], array) vwtrainflist - File names for training visual words vwtrainID - Class indicators for training visual words testflist - File names for test testID - Class indicators for test The class indicators are formed by -1 -1 -1 +1 -1 -1 -1 +1 -1 -1 -1 -1 -1 -1 : where the i-th sample belonging to the j-th class is indicated by the i-th row which has +1 only at the j-th column and 0 on others. The codes in "util" generates those 'cv_data_*.mat' files. Usage ===== First, you rearrange the image files in dataset as mentioned above into the directory (PATH_TO_IMAGES) and prepare the file of 'cv_data_*.mat' in the root directory (ROOTPATH). Then, by indicating the path to vlfeat that you have already installed, just type >> run_all After finishing all procedure, you can find the following directories in ROOTPATH /phow_desc/ : SIFT local descriptors /visualwords/ : visual words /grad_dir/ : Basis direction for coding orientations of p.d.f gradients /result/ : Classification results Reference ========= If you use this package, please cite it as T. Kobayashi, "BoF meets HOG: Feature Extraction based on Histograms of Oriented p.d.f Gradients for Image Classification," Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. -, 2013

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：