Classify of Image based on my pre-trained model-
How I have completed this ?
With the help of different knowledges resources and support, I have done below to make this completed
I have collected the sample Images of 4 individual class(Celebrity).
with managed filing and labeling system I cleand the data as per usefull images like images with full faces, proper 2 eyes.
In OpenCV library I have utilised the eyes and face harcascades classifier to classify the face and eyes in each images. I further made it croped to train and collect the data for further processing.
To make the images easily understandable to machine I have used the wavelet image transformation for each cropped images.
I further catagorised my class to number to name and name to number.
Now as its the time for reshaping, combining the coloured croped with wavelet called Vstacking and further training combined images to make a model with the algorithms in gridsearch behaviour. And get the best result as per training and testing refrences.
from here I saved the model to pickle file and further implimented with my flask server. But here I am implimenting with Django with same attributes. Here it will take data from user, make the processing of images as before making model, and finally predict with model and gives the best result on available class and identify the class.
for hosting in web I have choosen my cloud VPS where I have maintend same environment as of production and make apache to serve the page. As it is of python here is role of wisgi for both django and flask. Working with flask is easy. But I also want to test with Django. So this is now running in Django
This is my Research project that I have done solely with the refrence of online materials and the youtube. My teachers Mr. Redda Dehak, Louriq Fataha, Daval have given very good Guidelines and motivations for the completions of this project. The algorithms and concept that they have taught me are very useful for this project.
This project is only for classifing the pre-trained model. Now I have only trained 4 class for my training model. With the help of Support Vector Machine Algorithm. I have implimented the gridSearch model to identify the the best outcoume form the different algortihm like SVM, Random Forest and Logistic Regression. But the value of the SVM was high so I have choosen this SVM.
|Jupyter Notebook for Creating Model|
|Collection and Data Cleaning of Images for Classes|
|OpenCV with Harcascades|
|image wavelet transform, V-Stack, Training-Testing|
|Scikit -Learn for collection of Algorithm|
|Grid-Search for Best Result|
|VPS-Apache - Wisgi for Production|