Documentation
πŸ”— Nodes
πŸ—ƒοΈ Data

πŸ—ƒοΈ Data Nodes

Data Nodes interact with your file system. Whenever you read or write files, you should use these nodes.

πŸ“– Read

The Read Node is responsible for reading images from a directory. During processing, it goes through all the files within that directory. Currently, we support images in the formats: .jpg, .jpeg and .png.

Data Read Node
You can preview the images that are being read.

✍️ Write

Use this node whenever you want to save the resulting image to a file. You can define both the target directory as well as the prefix of the saved file.

Data Write Node
This node will save the input image in the outputs/ directory with the names result_[ID].png

πŸ“¦ Write Bounding Box

This node is useful whenever you generate a dataset for object detection task. This node calculates the bounding boxes for an image. The bounding boxes are calculated under the assumption that the background of the object of interest is transparent. In other words, if you want to save a bounding box for an image, make sure that your object of interest is the only object in an image and that the background is transparent.

Data Write Node
The saved bounding boxes will contain the fruit visible on the original image.

JSON

The bounding boxes can be saved in a JSON format in the following way.

[
    {
        'x': 10,    // Left 
        'y': 50,    // Top
        'w': 30,    // Width
        'h': 20,    // Height
        'x2': 40,   // Right
        'y2': 70,   // Bottom
    },
    ...
]

YOLO

Currently, Rodina Flow does not have support for multiple classes so an example output of a YOLO-formatted bounding boxess could be the following.

// class x_center y_center width height
0 0.270020 0.266113 0.133789 0.155273
Rodina FlowRodina Flow - 2024
XLinkedIn

Product

Documentation

Stay up to date

Join the waitlist