Algorithm of Scene Text Detection
### Author:Gen.Li ### E-mail:[email protected]
####1.Description:
this program show a demo of text detection using the Algorithm in
SCENE TEXT DETECTION WITH EXTREMAL REGION BASED CASCADED FILTERING
####2.Needed Materials (important):
(1)Models
Models are on : 3rd_party_tools\
broad_pca_data_Dec_06
pca_data of broad type
narrow_pca_data_Dec_06
pca_data of narrow type
broad_svmStruct_Dec_06
broad SVM
narrow_svmStruct_Dec_06
narrow SVM
detnet_layers
cnn model
(2) Matlab platform is needed. the code tests on Matlab 2014b
.
(3) You need a windows7 x64
system with CUDA 6.5
or a linux
system with CUDA7.0
.
if you have other version of CUDA on either system, you need to compile the Matconvnet
and get your own mex file, because the off-the-shelf model in [1]
is used, for more information, see the use of model in Deep features for text spotting
(3) data
you can use img_1.jpg
available or your own data , you need to change the path in single_img_pipeline.m
####3. Use
run single_img_pipeline.m
[1] M.Jaderberg, A.Vedaldi, and A.Zisserman, “Deep features for text spotting” in Proceeding of ECCV, 2014, pp. 512–528.