Congratulations for completing this course. It means a lot, because you have spent considerable amount of time going through each and every lecture. And I hope it was worth. So definitely this will arm you with a basic fundamental knowledge of how computers are machine processes and image and how you can tweak and train the model. Since specifically, we learned about the image basics where you saw the difference between the way human persons they made the machine perceive the image, you saw how much impressive the image as just a bunch of pixels. And you also saw the difference between a black and white image and the color image.

How there are three channels in case of color image, whether there's any one channel case of black and white image and you also saw how we flatten out the image of three channels into just one channel, so that it's easier for a machine to process it. Then you also learned about evaluation metrics like the metrics like accuracy how the accuracy fails in certain scenarios, and why we need to go for precision recall, and most importantly, the f1 score, which is the harmonic mean of precision because at the end of the day, we need to maximize the f1 score. And that's how we evaluate each model. Then you also learned about the open CV tool library, which is used for processing the image. And we also saw a class called preprocessor, where we use open CV two library to resize the image into 32 by 32 pixels.

We also saw how we can use open CV to load the image from the disk into an array of intensities. Right and that we can process them further. We learned about the K nearest neighbor algorithm how you use this the voting system to classify the nearest neighbor and you also varied its hyper parameter like number of neighbors to see the fact the evaluation matrix right. So, armed with this information, definitely you can apply a various terms like CNN in STM sorry, name, Guru, to classify your images, we kept the things very simple in this particular cause, so that we didn't want you to overwhelm with a lot of complexities of machine learning at this point. So, at this point, definitely you I feel you have a basic understanding of what is an image classification and what are the different steps that are required to classify an image, how you basically read an image into a NumPy array.

And then how you flatten the image, how you split the data into a training set and the test set in how you evaluate the matrix in your outer what are what are the matrix mean internally. So these knowledge is common throughout any machine learning concept to take any algorithm that you check take take advantage metric doesn't change open system library doesn't change the way you know or the image doesn't change. Only thing that thing is actual was that instead of KNN, you might be using CNN and an STM or guru, but rest of the steps will say, so, at least at this point, are you I would say you have learned at the 75% of ml only thing that you can take yourself to the next level is just by experimenting with a different golf of CNN and STM and then play with him and you will be become master of machine learning in no time.

So all the best practice the examples that have You win again and again, and keep experimenting with new algorithms and things. That's the only way you will develop understanding of machine learning. Thanks whenever I say bye