@@ -19,7 +19,7 @@ After preparing the required environment, you need to construct the bases which
Then, it builds the source and target domains, a domain being a set of 128x128 forged or not forged patches obtained from spliced images of DEFACTO (eventually compressed before the cutting into patches).
Please note that by default, each domain is made of 3 fold in order to realize a 3 fold cross validation. If you don't want to do a cross validation or if you want to test more fold, you can change it straightforwardly in the file `construct_domains.py`.
*To be sure that you construct the same domains as ours, we saved a .txt file containing the order of the image paths we extracted from the Splicing category. By default, the code available here take into account that list and not the one you could obtain from your side that can lead to a different order.*
*To be sure that you construct the same domains as ours, we saved a .txt file containing the order of the image paths we extracted from the Splicing category. By default, the code available here takes into account that list and not the one you could obtain from your side that can lead to a different order.*
- When this is finished, you can reproduce our experiments lauching the script `simulations.py` moving to the folder `Experiments`. By default it will reproduce all the experiments of the paper given that you created previously all the necessary domains.
...
...
@@ -48,7 +48,7 @@ Please find below a table linking a code to an experiment presented in the paper
- If you want to make other experiments feel free to precise your hyperparameters using a dictionary like the one in `.\Results\SrcOnly-s=none_t=qf(5)\hyperparameters-SrcOnly-s=none_t=qf(5)` before calling the function `simulate(hyperparameters)`.
- Each experiment leads to the creation of a folder in `Experiments\Results` where the hyperparameters you gave, the weights of the best models found during the trainings and the final results are stored progressively. This folder has a name deduced
from the filenames of the source and the target. For instance SrcOnly(None --> QF(5)) leads to the construction of the folder `SrcOnly_s=none_t=fq(5)`. You can tweak a bit this name using the key `'details'` in the dictionary `hyperparameters`.
from the filenames of the source and the target. For instance SrcOnly(None --> QF(5)) leads to the construction of the folder `SrcOnly_s=none_t=qf(5)`. You can tweak a bit this name using the key `'details'` in the dictionary `hyperparameters`.
- The name *hyperparameters* is in a broad sense since this dictionary should contain information such as learning rate, batch sizes, etc... but also the filenames (with extensions) of the bases you want to use as your source and your target.
...
...
@@ -111,7 +111,7 @@ from the filenames of the source and the target. For instance SrcOnly(None --> Q
