Pokemon_Computer_Vision_Project

This was a project that I started after getting my Google Data Analytics and DeepLearningAI TensorFlow Developer certificates from Coursera.

The main idea is this, use Kaggle to get access to Pokemon images and pokemon data. Merge the two datasets somehow, and then analyze for trends that may influence a ML model. After that create a model in TensorFlow to predict the first type from a pokemon based on the image provided.

With this project, I intended to demonstrate my skills in

1. data sourcing
2. data cleaning
3. data analytics
4. data visualization and storytelling
5. OOP
6. pandas
7. TensorFlow models
8. Computer vision in particular
9. File management in Google Colab

The project has 3 parts:

1. data sourcing and cleaning

   a) here I use Kaggle APIs to get access to two pokemon data sets

   b) I also merge and clean them to get them in a form that I want

2. data analysis

   a) here I use pandas to see if there are any trends in typing over time

   b) Then I use matplot to make some visualizations and explain any trends that might effect my analysis

3. machine learning

   a) I use file management, pandas, and google colab to preform a repeatably randomized train test split on my images

   b) I use OOP and develop a Python class called Pokemodel which has internal methods which preform all of the model's features

   c) I plot the accuracy vs. epochs to see overfitting and point out potential fixes

   d) I take a look at some of the mistakes of the model and try to see where it is going wrong

This was project was a fun way to spend some of my Spring break in 2023, and I hope that you enjoy it!