The project aims to test the existing neural network algorithm for image classification of Meme images such as muffin and chihuaha dogs which are very similar to each other.

-> Open your Google Drive

-> Upload the .ipynb file > Open with > Google Colaboratory

Edit > Notebook settings or Runtime>Change runtime type and select GPU as Hardware accelerator

Google Colab is built for running Machine Learning projects hence has most of the libraries installed, but for assurance we can copy paste the entire list of libraries below and install them at a go.

!pip install or !apt-get install

#Must install before running
!pip install colorama

#These libraries do exist in colab, but you can reinstall them if required
!pip3 install tensorflow==1.8
!pip3 install keras
!pip3 install torch torchvision
!apt-get install python-numpy python-scipy
!pip install os
!pip install math
!pip install shutil

At the very end of the notebook, GUI will automatically run for testing, here you can upload a meme image from your local server/desktop and check the accuracy. You can download the dataset from the repository.

You can run this block as many times you want to test the images. The prediction/result will be displayed below the image with the calculated probability.

The entire notebook will take 12-15 minutes to run all blocks in GPU environment.

• @AdityaShetty_21200138

• @TejaBhat_21201067