Model that able to distinguish picture between dog, cat, or panda.
Dataset were taken from Kaggle
How to run this code :
- Download the dataset from kaggle
- Upload the downloaded picture to your google drive
- Open the ipynb in Google Colab, else the running time will be very long
- Set the running type into GPU
- Change the file directory of data to where you save the dataset in your drive
- Run the code
Dataset contains 1000 pictures of cats, dogs, and pandas. Then I divided into train : test with test_size ratio of 0.2. Since I used google colab, I decided to greed it out and use colorfull picture, which is actually 3 times bigger than grey color picture.
This model were make using keras - tensorflow with Sequential model
Here is the summary :
Confusion matrix :
This model is able to predict with up to 73% accuracy.
The problem is this model was not so smart when differentiating cat and dog.
But, since I choose color picture, it is able to detect panda better than the other.
As I will keep learning about deep learning, I believe this model's accuracy STILL can be improved.