-
KNNFeatureExtractor.py: This is used for reading all training images from the specified directory and extracting features from them. The feature sets are stored in the specified csv file.
-
KNNTrainer.ipynb: This is used for training KNN Classifier using the feature sets stored in the specified csv file.
-
HMMFeaturesExtractor.py: This is used for reading all training videos from the specified directory and extracting hand pose, motion features from them. These feature sets are stored in the specified csv file.
-
HMMTrainer.py: This is used for training the HMM Classifier using the feature sets stored in the specified csv file.
-
merger.py: This is used for copying images from one folder to another. This is done by counting no. of images in destination folder, then renaming the source images starting from this count. Thus, the source images don't replace destination images because of renaming.
The sample csv files have been provided so that reader can see how the feature set csv files are generated. All these sample csv files are stored in SampleFeatureSetFiles folder. This currently has these sample files:
-
silatra_digits_letters_pregenerated_sample.csv: This has the feature sets generated by reading all letters and digits training images for training KNN.
-
silatra_gesture_signs_pregenerated_sample.csv: This has the feature sets generated by reading all intermediate gesture signs' training images for training KNN.
-
silatra_gestures_pregenerated_sample.csv: This has the feature sets generated by reading all gestures training videos for training HMM.
The sample KNN and HMM Models have just been included for reference, to see how they would look like after training and saving the models.
When you try to train KNN or HMM models or try to generate Feature sets, they will be stored in Temp_* folders. This is done to prevent ruining the sample folders. If you want to use it properly, you need to understand the code so as to understand which variables to modify to save the models properly.