Merge branch 'main' of https://github.com/jsten07/CNNvsTransformer in…

…to main

README.md

@@ -1,35 +1,50 @@
-# CNNvsTransformerComparing CNN- and Transformer-Based Deep Learning Models for Semantic Segmentation of Remote Sensing Images
-Master Thesis at the Institute for Geoinformatics, University of Münster, Germany. 
+# CNNvsTransformer: Comparing CNN- and Transformer-Based Deep Learning Models for Semantic Segmentation of Remote Sensing Images
+Master Thesis at the Institute for Geoinformatics, University of Münster, Germany.
 
 ### Description
-Dataset used: [ISPRS Benchmark on Semantic Labeling](https://www.isprs.org/education/benchmarks/UrbanSemLab/default.aspx) (just used Potsdam dataset so far)
 
-### How To
+With the given code, results for the thesis mentioned above were created.
+A pdf of the thesis will be added soon.
 
-1. Prepare dataset in following folder structure:
+In the following some basic steps to use the code are described but might need some adaptations, based on the utilized infrastructure.
+
+Further documentation will be added soon. 
+
+### How To Train the Models
+1. Prepare data
+	1. Download data, for example one of:
+		1. [ISPRS Benchmark on Semantic Labeling](https://www.isprs.org/education/benchmarks/UrbanSemLab/default.aspx)
+		2. [FloodNet](https://github.com/BinaLab/FloodNet-Challenge-EARTHVISION2021)
+	2. Patchify data into appropriate sizes, e.g. $512\times 512$
+		1. `\helper\patchify.py` might be useful
+	3. Split data into train and test data in the following folder structure, with `rgb` folders containing the corresponding images to the groundtruth labels in the `label` folders:
 ```
 |-- data
 |   |-- rgb
-|   |-- rgb_valid
+|   |-- rgb_test
 |   |-- label
-|   |-- label_valid
+|   |-- label_test
 ```
- - with ground truth images in folders `/label` having the same name as the corresponding images in folders `/rgb`.
- - Ideally also add folder `rgb_test` and `label_test` and put respective data in there by splitting all the data into train, validation and test data. 
-
-2. Install requirements (compare `/PALMA/requirements.txt` and modules loaded in `/PALMA/train_unet.sh`; respective requirements file will be added later).
-3. Train model with
+2. If you use the PALMA cluster, just adapt and run one of `\PALMA\train_unet.sh` or `\PALMA\train_segformer.sh` and you are done
+3. Otherwise: Install requirements (see `\PALMA\requirements.txt` and modules in `\PALMA\train_unet.sh`) 
+4. Look at possible parameters in `train.py` and run the following line with respective adjustments:
 ```
-python3 train.py --data_path /your/path/to/folder/data --output_path ./weights
+python3 train.py --data_path /your/path/to/folder/data --name ./weights
 ```
+5. Find your model in folder `./weights`
+
+By default a U-Net model will be trained for 20 epochs. Further default settings can be derived from the parameters of `train.py`.
 
- - By default a U-Net (alternative: `--model segformer`) is trained with 20 epochs (alternative: `--epochs integer`).
+### Evaluation and Visualization
 
- - Furhter script arguments are listed in `/train.py` and can be shown with `python3 train.py --help`. 
+The evaluation and visualization of the models was done with help of the notebooks in the respective `./Notebooks` directory.
 
-4. Find your model weights in folder `./weights`.
+- `compare.ipynb`: comparison of two models by visualizing predictions of both and calculating metrics on test data
+- `homogeneity.ipynb`: calculation of clustering evaluation measures
+- `radarchart.ipynb`: plot radar charts and bar charts for the calculated metrics
+- `count_classes.ipynb`: count pixels per class and mean and standard deviation in an image dataset
+- OUTDATED: `Segformer_Run.ipynb`, `UNet_Run.ipynb`, `Segformer_visualize_attention.ipynb`
 
-5. Visualize and compare your models using the respective Notebooks in `/Notebooks` by changing the model paths and names and the data path.
 
 ### Acknowledgements