Memes_vs_Notes

All of us have thousands of pictures on our phones,
two major contenders for storage space are Memes and Notes

The Mobile App

The mobile app separates/extracts the notes from a folder specified by the user.

Creating Dataset

Keep the Creating Dataset Script in the same folder as Memes and Notes (named as 'Memes' and 'Notes' respectively) and run it.
This will produce numpy arrays MVN_X.npy and MVN_Y.npy which are images and labels respectively.

Architecture (Conv)

F - Filter, S - Stride, C - Channels
Conv Layer - F(11x11) S(4x4) C(3,96)
Conv Layer - F(5x5) S(4x4) C(96,134)
Max Pool - F(2x2) S(2x2)
Conv Layer - F(3x3) C(134,158)
Max Pool - F(2x2) S(2x2)
Dense - C(158x2x2,252)
Dense - C(252,1)

Optimizer - ADAM
learning_rate = 0.001
The Best dataset is saved as MVN2.pth

Accuracy Training Set (1500/1600) - 97.8%
Test Set (100/1600) - 100%

Architecture (Non-Conv)

Dense Layer - C(256x256x3,1000)
Dense Layer - C(1000,50)
Dense Layer - C(50,1)

Optimizer - ADAM learning_rate = 0.001

Accuracy Training Set (1500/1600) - 88.26%
Test Set (100/1600) - 88%

Data

Data is at https://drive.google.com/drive/folders/1SH5Y8fpwue2uIN1kOeyj_HweW4AP1xKO?usp=sharing
paste MVN_X.npy and MVN_Y.npy in the data folder

To open the Colab Notebooks

Goto - https://colab.research.google.com/github/ and then copy and paste the notebook links from the repository