Page 338 - AI Ver 3.0 class 10_Flipbook
P. 338
Stage 2 Data Acquisition
Relevant datasets are identified, collected, and prepared for analysis. For the coral bleaching project, the dataset
used is from the manuscript titled "Bag of Features (BoF) Based Deep Learning Framework for Bleached
Corals Detection." This dataset is intended for research and experimentation to develop AI models for detecting
bleached corals.
You can download the Coral Bleaching dataset from the following link:
https://orangewebsupport.co.in/AI/Coral_Bleaching.zip
OR
https://bit.ly/orange_computer_vision
Note, Image Analytics can be added by selecting Image option from Add-ins from Options tab.
The steps for acquiring training dataset are as follows:
Step 1 Open Orange Data Mining Tool and locate the Import Image widget.
Step 2 Drag and drop this widget onto your workflow canvas.
Step 3 Right-click on the Import Image widget on the canvas.
Step 4 Select the Rename option.
Step 5 Enter the new name: Training Data to make it easier to
identify the widget's purpose.
Step 6 Double-click the renamed Training Data icon to open
its configuration window.
Step 7 Click on the Browse button.
The Select Top Level Directory dialog box appears.
Step 8 Navigate the location containing your training dataset.
Step 9 Click on the Select Folder button.
Ensure that your dataset is organised into subfolders, where each subfolder corresponds to a category
or class (e.g., "Bleached" and "Unbleached").
Orange will automatically recognise the 2 categories based
on the subfolder structure in the training dataset. These
categories represent the 2 classes of the problem:
• Bleached: Images of corals that are bleached.
• Unbleached: Images of healthy corals.
336 Touchpad Artificial Intelligence (Ver. 3.0)-X

