## Dataset

The Food-101 dataset, originally made available at https://data.vision.ee.ethz.ch/cvl/datasets_extra/food-101/ is a data set of 101 food categories, with 101'000 images.

## Training
I trained this classifier with a Efficient B2 model, with just change last layers of this cnn.
The final results of the training  in 5 epochs  are as the below table.


100% ..........................................................5/5 [1:52:48<00:00, 1350.89s/it]


Epoch 0 | Train Loss: 3.1171 | Train Accuracy: 37.15 | Test Loss: 2.2336 | Test Accuracy: 59.80


Epoch 1 | Train Loss: 2.8259 | Train Accuracy: 43.46 | Test Loss: 2.1552 | Test Accuracy: 61.88 


Epoch 2 | Train Loss: 2.8085 | Train Accuracy: 44.20 | Test Loss: 2.1142 | Test Accuracy: 62.59 


Epoch 3 | Train Loss: 2.8071 | Train Accuracy: 44.47 | Test Loss: 2.1027 | Test Accuracy: 62.99 


Epoch 4 | Train Loss: 2.8082 | Train Accuracy: 44.43 | Test Loss: 2.0819 | Test Accuracy: 63.48 



Process Time: 6768.90 seconds

## Examples
The example images provided with random choices from test folders