Spaces:
Build error
Build error
File size: 1,065 Bytes
7a8fb38 07885c2 7a8fb38 07885c2 7a8fb38 07885c2 7a8fb38 07885c2 7a8fb38 07885c2 7a8fb38 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 |
## 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The Food-101 dataset, originally made available at https://data.vision.ee.ethz.ch/cvl/datasets_extra/food-101/ is a data set of 101 food categories, with 101'000 images.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 |