In this pipeline, I choosed EfficientNet_B0 and EfficientNet_B2 to train them with 2 datasets and different amounts of epochs to see what is the best combination to get the most accuracy for my task which is a food classification.
The instracture of using this pipline is in the main_(notebook_version).ipynb file. If you have any suggestion to upgrade this pipeline, feel free to open a pull request or let me know in the discussions page
Note: I used two datasets which is gathered from food 101 dataset. One of them has %10 and the other has %20 of the images of food 101 dataset. I did this to see how much the accuracy will get effect if we train the model with:
- different amount of data
- different size of model
- different train time (epochs)
The output of this pipeline will be shown in tensorboard, so we can see which model with which parameters have better accuracy.