What About a 6-Week Machine Learning Project? Beginners Friendly Cat vs Dog Classification Problem.

Source: https://www.sciencealert.com/when-it-comes-to-dog-vs-cat-brains-science-might-have-found-a-clear-winner

By enrolling for a Machine Learning online course, I got a chance to gain a foothold in the field of ML. Through this course, I came to know about Pandas, Numpy, Scikit-Learn, Tensorflow etc. All these technologies were initially a new thing to me, my starting curve being a basic knowledge in Python language. I got to know about the Jupyter Notebook, Google Colab which very much go in handy while dealing with Machine Learning frameworks.

One of the coolest stuff that I came across during this process of learning is a Deep Learning project to classify dogs based on their breed. It is an offbeat technique, which encompasses all the different steps that deal with the images while we are using them for classification purposes. These steps include:

1. Getting the workspace ready.
2. Getting images and labels.(Images obtained from Kaggle).
3. Turning images into Tensors.
4. Building the model.
5. Creating callbacks.
6. Training our model.
7. Saving and Loading our model.
8. Making predictions on the test data set.
9. Making predictions for the custom images.

The Github link is given here.

Dog Breed Classification Result On Custom Image.

So after the completion of this project, the spark of an idea fitted in my mind,that is - why shouldn’t I try applying all these steps to create a model on my own by searching through the documentation, trying out different things, asking doubts , attempting new models and so on.

I decided to stick on with a binary classification ML work to classify between Dogs and Cats. These kinds of work are readily available, being common classification problem that can be obtained on the internet. But this is also the kind of work that a newbie can ideally start with.

I am planning a 6 week timeline for the completion of this project which got kickstarted on 27th July. As a first step taken for likewise Machine Learning projects, I collected the required data set from Kaggle and uploaded it into Google Drive. My plan for this week is as follows:

1. Getting images and labels.
2. Creating our own validation set.
3. Pre-processing images.
4. Turning data into batches.

These are the tasks that I plan to compete during 1st week and am confident enough that I can carry it along diligently.

Weekly Task (Created using draw.io)

I am feeling very excited and positive about this project and even if I’m unable to complete this work in 6 weeks, I will still be happy that I have put my brain and best effort into it, which is indeed a great thing as doing something is better than nothing. So that’s all about my week 1’s preparation and still plenty of work is remaining to be completed this week.