0

I've been assigned to read text from captcha images. Input images training images are given. The part where I'm stuck is I've to train SVM (with that given training sample data). AFter a hell lot of search and headache, I still don't know how to start SVM training.I've install openCV from NuGet, EMGU CV and accord.net but still fruitless. Any help would be really appreciated. Captcha Images are like this. This is what my instructor step by step told: 1. There is Captcha data in the test folders. Divide the test data into two parts, training data (60%) and test data (40%) 2. Figure out a way to segment out the 4 letters in each image of training data. 3. Figure out a way to store them automatically in Class folders (all As in folder named “A” etc) 4. Train a SVM on the segmented data to get a training file. 5. Use the Test data with the trained file to read the captchas. 6. Report accuracy automatically. I'm badly stuck at step 4.

Chili Mili
  • 79
  • 1
  • 6
  • Your instructor is a bad person, captcha images are intentionally messed up to avoid OCR recoginition... – Gusman May 13 '16 at 15:03
  • Ok, look at this:http://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_ml/py_svm/py_svm_opencv/py_svm_opencv.html it's Python, but the concept would be the same, there you can see how OpenCV is trained. Hope it helps, and give my salutation to your instructor XD – Gusman May 13 '16 at 15:07
  • Can you share the training data ?? :p I would like to try :p – Arijit May 13 '16 at 15:09
  • I do have very little few concepts, but don't know how to implement them. Looked at the the following post too http://stackoverflow.com/questions/14694810/using-opencv-and-svm-with-images – Chili Mili May 13 '16 at 15:11
  • Sure @ArijitMukherjee, how do i send ? – Chili Mili May 13 '16 at 15:12
  • google drive or dropbox – Arijit May 13 '16 at 15:15
  • https://drive.google.com/open?id=0B76GzPjQEDTzejJkNFNidUdfYTQ – Chili Mili May 13 '16 at 15:31

0 Answers0