Category Recognition

This tool allows image classification into different categories using OpenCV and Machine Learning techniques.

Dependencies

• OpenCV

Resources Structure

• resources/
• random_images/
• categories_train/
  • category_1/
  • category_2/
  • category_3/
  • ...
• categories_test/
  • category_1/
  • category_2/
  • category_3/
  • ...

Run

1. Fork, clone or download the source from GitHub
2. Compile it using make
3. Create vocabulary file with a bunch of random images ./make_vocabulary <imgaes_folder_path>
4. Make a folder to add categories folders with images to train SVM
5. Create Bag of Words descriptors of your categories train images ./bow <vocabulary_file_path> <train_images_folder>
6. Now you have a file with the trained SVM and in console output you have the labels.
7. Classify with ./classify <vocabulary_file_path> <SVM_path> <new_image_path>". It will return a number you must match in the computed labels when computing Bag of Words