Classification of beer labels using Deep Learning (Keras) bringing into focus low inference time.

This is a brief research about beer label classification from images focusing on low inference time.

The dataset used contains five categories of beer brands.

How to run

There are 3 files in total. For the files that have the ending "_colab.py" it is suggested to be executed on the Google Colaboratory platform, this way, there will be no problems with dependencies.

Regarding the file of inference time test that runs on local machine, it is necessary to install the tensorflow package, which will install keras automatically.

pip install tensorflow

Also, the dataset folder must be organized this way: (Pay attention to the file paths that are in the codes)

dataset
├── test
│   ├── Becks
│   ├── Brahma
│   ├── BudWeiser
│   ├── Corona
│   └── Skol
├── train
│   ├── Becks
│   ├── Brahma
│   ├── BudWeiser
│   ├── Corona
│   └── Skol
└── validation
    ├── Becks
    ├── Brahma
    ├── BudWeiser
    ├── Corona
    └── Skol

It is suggested to split the training, testing and validation data into 60%, 20% and 20% respectively.

It is possible to run the algorithm in the notebook step by step by splitting the code to each blank line following the same sequence.