Made flowchart better

Author: TaqiHamoda

JOSS/flowchart.png

@@ -83,7 +83,7 @@ This project implements this entire workflow in a configurable and automated pip
 
 # Methodology
 
-![The implemented classification framework is depicted in the above flowchart. The left section outlines the primary stages of the pipeline, while the right section details the specific components and features utilized in each stage. \label{fig:flowchart}](flowchart.png)
+![The implemented classification framework is depicted in the above flowchart. The left section outlines the primary stages of the pipeline, while the right section details the specific components and features utilized in each stage. The dashed arrows represent the path being optional. \label{fig:flowchart}](flowchart.png)
 
 The `Thematic2.5D` pipeline processes images, depth-maps, and masks to generate a labeled dataset of uniformly sized tiles for training and testing. It identifies potential object and background locations using masks, extracting corresponding square tiles from image and depth data. To address class imbalance, a fixed number of background tiles are sampled per image, and only a subset of available object pixels are used as tile centers. Data augmentation is applied by rotating object tiles at multiple angles.