Update README.md

README.md

@@ -8,3 +8,44 @@
 
 
 ## Step 1: Annotating the parts using LabelImg
+
+![alt text](/fig/fig2_labelImg.png)
+
+See in the image below how the folders of each class are organized.
+
+![alt text](/fig/fig3_annotation.png)
+
+Each folder is a class. In the folder (class) the images are annotated with bounding boxes around the component parts of each of the insects. Each image is associated with an XML file. 
+
+The dataset is structured as follows:
+
+![alt text](/fig/fig4.png)
+
+The Pests4C2P dataset can be accessed at: [https://doi.org/10.6084/m9.figshare.20055602.v1](https://doi.org/10.6084/m9.figshare.20055602.v1)
+
+## Step 2: Create the samples for training/validation/testing using cross-validation.
+
+### 1) Clone the project
+git clone http://git.inovisao.ucdb.br/inovisao/crossvalidation-repetition-object-detection.git
+
+Put Pests4C2P dataset folder on same folder level as crossvalidation-repetition-object-detection folder.
+
+![alt text](/fig/fig5.png)
+
+### 2) run the project to generate the samples
+
+    cd crossvalidation-repetition-object-detection/
+
+    python3 main.py ../Pests4C2P ../fold cross 5 True 0 True
+
+Where:
+
+- _**../Pests4C2P**_ - the folder where the dataset is; 
+- _**../fold**_ - the folder where the folds will be generated (folder is created automatically); 
+- _**cross**_ - it indicates that it will be cross validation; 
+- _**5**_ - number of folds (only 5 or 10); 
+- _**True**_ - if there will be the validation set; 
+- _**0**_ - percentage of data increase; 
+- _**True**_ - if there will be oversampling;
+
+