No description, website, or topics provided.
Switch branches/tags
Nothing to show
Clone or download
Pull request Compare This branch is 39 commits ahead of kyledemeule:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Note: Extracted features files can be found in extracted_features/, classification can be done directly on those.

Processing Images

To process raw images (i.e. crop, rotate, and label), run this command:

python raw_images/ processed_images/

Raw images should be in the raw_images/ folder. Within that folder there should be a folder for each class containing all the images belonging to that class. Processed images will be saved in the second argument, processed_images/ in this case, along with a classes.txt file (dictionary containing the index to class mapping), as well as an item-labels.txt file (another dictionary containing the image filename to class index mapping).

Extracting Features

To extract features using a neural network, make sure you have the alexnet model installed and in a caffe/ directory adjacent to this project folder. You can customize that info in the file. If you don't have the necessary model, the file will tell you what you're missing. run this command:

python fc8 processed_images/ features_fc8.npz

First command is where to get the features from the neural net. Second is directory of processed images, and third is where to save the result. Make sure the last file ends in .npz


To run classification on the extracted features call:

python extracted_features/features_fc8.npz

The file you pass in is the features from the previous step (or use one of the pre-extracted ones in extracted_features/).