Skip to content

The project aims to create a Supervised Machine Learning algorithm for cat classification problem that detects whether a cat is present in the image or not by generating an RGB/HSV Histogram Model. By processing a large set of images, the specific features are being identified with an accuracy of up to 83%.

Notifications You must be signed in to change notification settings

mirunamariafatu/Cat-Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 

Repository files navigation

Fatu Miruna-Maria
311CA

Part 3 - Householder Prediction

RgbHistogram:
	- se construiesc matricile red, green, blue extrase din img
	- cu ajutorul functiei idivide acestea sunt scalate la intervalul dorit
	- intrucat idivide va returna si 0, iar functia accumarray primeste pozitiile elementelor (pozitia 0 nu exista), adaugam +1
	- se creeaza histogramele care vor fi adaugate in sol
HsvHistogram:
	- fiecare pixel din r,g si b va fi transformat in h, s, v
	- analog RgbHistogram
SSt si Householder:
	- urmeaza algoritmul prezentat la seminarii
Preprocess:
	- mai intai sunt extrase histogramele pozelor cu pisici, apoi cele fara pisici
	- matricea X contine toate histogramele pozelor, iar y etichetele pozelor (1 sau -1)
Learn si Evaluate:
	- evaluate este construita asemanator cu preprocess
	- se obtine w si se construieste noul y
	- variabila corect creste de fiecare data cand conditia de poza cu pisici/fara pisici este indeplinita

Part 4 - Gradient Descent Prediction

Learn :
	- scalez matricea X si adaug coloana finala de 1
	- generez vectorul coloana w dupa algoritmul Mini-batch Gradient Descent
Evaluate:
	- cu functia preprocess generez toate histogramele imaginilor in matricea X
	- scalez matricea X si adaug coloana finala de 1
	- construiesc noul y
	- pentru fiecare set de poze(cu pisici si fara pisici) verific daca acestea au fost corect identificate, in caz afirmativ variabila corect este incrementata
	- calculez procentajul imaginilor corect identificate
	

	


About

The project aims to create a Supervised Machine Learning algorithm for cat classification problem that detects whether a cat is present in the image or not by generating an RGB/HSV Histogram Model. By processing a large set of images, the specific features are being identified with an accuracy of up to 83%.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages