Here I used transfer learning to generate new recepi. The dataset consists of a lot of basic food items like egg, soup, bread etc. I also performed some image augmentation to increase the no of training dataset. I used the trained model InceptionResNetV2 which has almost 54M parameter. Then I add some other layer with this model and train 43k parameter. The training will take a lot of time. Thats why after training I commented out the cells.
The notebook contains a lot of theory that is needed for the model. Also you can check the image folder for the theory.
Source code for python is added with .ipynb format
prabormukherjee/Transfer_learning_food-Classification
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|