No self-introduction
Loading
Profile
Follow
codes (6)
Text detection via DCT energy
no vote
In [1], the DCT coefficients of a group of intra coded blocks are calculated as texture measure, and the energy measure is used to describe the original method. The key technology energy calculation code is provided by the text detection baseline algorithm [2]. [2] R analysis, D goldgof, P. soundarajan, V. Manohar, J. lovelaw, R. Bowles, M. Boonstra, V. korova, and J. Zhang, "face, text of performance evaluation framework, and video detection and tracking of vehicles: data, indicators, and protocols," IEEE model analysis and machine intelligence, Vol. 31, No. 2, P. 319? 36，2009。
lzd0825 0 0
1
Detect text with stroke width transform
no vote
Detect text with stroke width transform.Dependencies : OpenCV, boost.To compile:g++ -o DetectText TextDetection.cpp FeaturesMain.cpp -lopencv_ core -lopencv_ highgui -lopencv_ imgproc -I/path/to/current/directorywhere /path/to/current/directory is replaced with the absolute path to the current directory.To run:./TextDetection input_ file output_ file dark_ on_ lightwhere dark_ on_ light is 1 or 0, indicating whether the text is darker or lighter than the background.
lzd0825 0 0
1
Text Detection by DCT energy localization
no vote
The original method described in DCT energy localization text detection [1] is used as a set of energy-saving measures to calculate the DCT coefficients of the intra coding block of the texture measure. The energy calculation code comes from the text detection baseline algorithm provided in [2]. [1] N. Chadha, R. Sharma, A. Agra and A. Gupta. &In the 28th islomar conference, signals, systems and computers, 1356-1361, October 1994. [2] R. Kasturi, D. goldgof, P. soundarajan, V. mano, J. garofolo, R. Bowles, mishochang, V. korova, and J. Zhang & quot; video performance evaluation framework for face, text, and vehicle detection and tracking: data, metrics, and protocols & quot; IEEE transaction pattern analysis and machine intelligence, Vol. 31, No. 2, pp. 319-336, 2009.
lzd0825 0 0
1
Caffe training and learning
4.0
Caffe study notes 4_ learn to build your own network--MNIST training and learning learning notes at the Caffe 1_CIFAR-10 2_ training and learning at the Caffe notes using a training model to extract characteristic study notes 5_ visualization of layer-by-layer image features
lzd0825 0 0
1
K-nearest neighbor classification
no vote
1. calculation data to be disaggregated data and each in a different class of distance (Euclidean or Mahalanobis).
2. choose the smallest k data from before, there used to select sort method.
3. compare the k distance before, find k data contains most of the classes of data, namely for categories containing class.
lzd0825 2016-08-23
lzd0825 1 0
1
Application of knn algorithm in statistical learni
no vote
Question:
lzd0825 0 0
1
No more~
Follows0
Fans0