How and Why I built it ?
I liked Pokemon since my childhood and I grew up watching them on TV, collecting cards and playing games. By the end of my graduation I got fond of getting into Machine Learning a bit. So I decided to train my own CNN on the Pokemon images and built my own Model to predict them correctly. In short, my own Pokedex in a python script :)
So, I studied about CNNs, Computer Vision, tensorflow and along with that, I got to learn Bing Image Search API to download a lot of images for my dataset. I trained it for 5 Pokemon classes, Yes the popular ones: Pikachu, Bulbasaur, Squirtle, Charmandar and Mewtwo! Though The model is quite generic and can be used to train on all Pokemons provided enough dataset.
And here we go! my own Pokedex to detect the Pokemon from an image.
Top comments (0)