readme.md

Python Image Clustering (Unsupervised Classifier)

Command Line Arguments

-i PATHNAME	# Input directory
-o FILENAME	# Output .PKL file
-p NUMBER	# Number of processes for pre-processing
-s True/False	# Whether to save output and graphs from HOG

Example Command Line Statements

python3 main.py 		# Runs the classifier with default settings
python3 main.py -p 16		# Runs the classifier with the specified number of 					     processes. 1-63 are valid inputs
python3 main.py -s True		# Saves the outputs of HOG and plots their feature 					     graphs

Non-Standard Libraries Used:

matplotlib      3.5.2
opencv-python	4.5.3.56
scikit-image	0.19.3
scikit-learn	1.1.1
scipy	       	1.8.1
threadpoolctl	3.1.0
tqdm	        4.64.0
imutils 	0.5.4

Outputs

/temp/*.pkl - These files are temporary and can be ignored. They are removed at the 			  beginning of each run
/plots/* - The feature plots of images (in no particular order)
/processed/* - The visualizations of HOG data (in no particular order)
output.pkl - The consolidated data matrix after pre-processing
2d_representation_plot.png - 2d State-Space graph of data
Elbow_graph.png - Graph of # clusters vs variance