================================================ Image Recognition - Python-OpenCV Implementation

• Image Recognition using SIFT and BRISK feature detectors

Installation

• We are using the latest opencv library version 3.0.0rc1
• Run pip install -r requirements.txt
• Test running it by python beyondbagsoffeatures.py

Dependencies

• Scipy
• Numpy
• scikit-learn
• matplotlib

Test Data Set

• The image data set need to run the code is provided in the images folder. (User may change it as per requirement.)
• Currently there are 3 datasets namely

• training and testing form the 1st pair
• caltech_train and caltect_test form the 2nd pair
• c1_train and c1_test form the last pair

• User can change the dataset one wants to run by commenting the line#30-37 of beyondbagsoffeatures.py.
• User can switch between SIFT and BRISK by setting sift=False on line#28

Results

References

• S. Lazebnik, C. Schmid, and J. Ponce, “Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories”, IEEE CVPR, 2006
• https://github.com/wihoho/Image-Recognition