ROI Operation Guide
1. ROI Introduction
ROI, short for Region Of Interest (ROI), refers in the machine vision field to the region selected from the target image that requires focused processing.
ROI is divided into two categories. One is ROI 2D (2D ROI), which is a region selected on a two-dimensional plane and considers only the two dimensions of length and width. It is usually a rectangular, circular, or other shaped region on an image or screen, similar to marking out a specific area on a sheet of paper.
The other is ROI 3D (3D ROI), which is a region selected in three-dimensional space. In addition to length and width, depth must also be considered. It can be imagined as defining a spatial area in the real world.
ROI has three functions:
Improve processing efficiency: When processing image data with relatively high resolution, analyzing the entire image often requires a large amount of computing resources and time. By specifying an ROI, processing can be limited to a specific area, greatly reducing the computation load.
Enhance analysis accuracy: ROI is the most critical specific area in an image. Through fine-grained analysis of the ROI, the detection, recognition, and measurement accuracy of specific targets can be improved.
Eliminate interference factors: An image may contain regions unrelated to the analysis target, and these regions may affect the accuracy of the analysis result. By defining the ROI, such interference can be excluded, improving the reliability of the analysis.
2. Set ROI
In PickWiz, ROI settings include ROI 2D and ROI 3D settings.
2.1 Add ROI
（1）Click ROI in Runtime Configuration, then click +Add ROI to add an ROI configuration. The interface will automatically switch to the ROI configuration page, where you can configure the new ROI.
（2）You can edit the ROI configuration name manually. Chinese characters and uppercase/lowercase English letters are supported, but punctuation is not supported.
（3）Before setting the ROI, you need to load the target image. There are two loading methods:
- Click
Capture Image, and PickWiz will trigger the Camera to capture an image and automatically acquire the 2D image and 3D Point Cloud.
- Click
Import ImageorImport Point Cloudto manually select the target image from a folder and load it into PickWiz. ClickImport Imageto import a 2D image for configuring ROI 2D, and clickImport Point Cloudto import a 3D Point Cloud for configuring ROI 3D.
（4）After loading the target image, you can set ROI 2D/3D.
The left side of the visualization window is the 3D Point Cloud window, and the right side is the 2D image window. You can drag to adjust the sizes of the 3D Point Cloud window and the 2D image window, as shown below.
2.2 Set ROI 2D
（1）ROI 2D Selection is used to select a specific ROI 2D for editing;
ROI 2D Name can be modified and supports Chinese characters and uppercase/lowercase English letters;
ROI 2D Type includes Workspace, Container/Pallet/Base Tray, and Observation Area. Each ROI 2D must have a corresponding ROI 2D type selected. PickWiz performs specific processing and analysis for each type area.
- Workspace: Mainly covers the reachable working area of the Robot and vision system, and is the most commonly used type area.
Note
During the process of adding an ROI 2D, the system will add a Workspace by default, and each ROI 2D must have one Workspace set.
Container/Pallet/Base Tray: Mainly covers objects such as containers, pallets, and base trays at the bottom of materials, helping with subsequent recognition or differentiation of containers, pallets, base trays, and similar objects.
Observation Area: Mainly used in binocular scenarios. For details, see KINGFISHER Series Camera User Manual
（2）In the 2D image window, left-click and manually drag to create a quadrilateral ROI 2D box. Make sure the ROI 2D box contains the target area, which is the selected ROI 2D type.
（3）To adjust the size of the ROI 2D box, click and hold the left mouse button and drag any of the four edges or four corners of the quadrilateral box;
To move the ROI 2D box, select any area inside the ROI 2D box, then click and hold the left mouse button and drag.
（4）To delete the ROI 2D box, left-click any area in the 2D image window, then press the Backspace key on the keyboard.
（5）Click Save to save the current ROI 2D configuration.
（6）If you need to set multiple ROI 2D type areas, click + to add a new ROI 2D configuration; conversely, click - to delete the current ROI 2D configuration.
Each time a new ROI 2D configuration is added, the system names the ROI 2D by default according to the addition sequence. Custom ROI 2D names are also supported. Then set the ROI 2D according to the steps above.
Note
Under the same ROI configuration, when multiple ROI 2D items are set, the ROI 2D types cannot be duplicated. For example, ROI2D_1 and ROI2D_2 cannot both be set to Workspace.
2.3 Set ROI 3D
（1）ROI 3D Selection is used to select a specific ROI 3D for editing;
Workspace: Mainly covers the reachable working area of the Robot and vision system, and is the most commonly used type area.
Note
During the process of adding ROI 3D, the system will add a Workspace by default, and each ROI 3D must have one Workspace set.
Container/Pallet/Base Tray: Mainly covers objects such as containers, pallets, and base trays at the bottom of materials, helping with subsequent recognition or differentiation of containers, pallets, base trays, and similar objects.
Picking Area: Mainly covers the area where the Pick Point of the Target Object recognized by the vision system is located. In general, the Picking Area is smaller than the Workspace. The purpose of setting a Picking Area is to resolve certain pose issues in depalletizing scenarios or ordered loading and unloading scenarios, and to avoid collisions caused by recognition errors.
Observation Area: Mainly covers areas that are frequently observed and monitored.
（2）After importing the Point Cloud, PickWiz automatically generates an ROI 3D box on the target Point Cloud. The center of the ROI 3D box is near the center of the target Point Cloud and can be fine-tuned in two ways:
- Manually drag in the 3D Point Cloud window
Place the cursor in the 3D Point Cloud window and scroll the mouse wheel up or down to adjust the display size of the ROI 3D box, as shown below.
Place the cursor in the 3D Point Cloud window, click and hold the left mouse button, and move the cursor to adjust the viewing angle of the ROI 3D box, as shown below.
Place the cursor in the 3D Point Cloud window, click and hold the right mouse button, and drag the yellow marker to adjust the size of the ROI 3D box. The ROI 3D box should fully enclose the Target Object and only needs to be slightly larger than the Target Object, as shown below.
Click and hold the right mouse button and drag the red/green/blue arcs to rotate the ROI 3D box. Red represents the X-axis, green represents the Y-axis, and blue represents the Z-axis. When rotating the ROI 3D box by dragging the arcs, the rotation angle value is displayed in real time.
- Enter values
Set the pose of the center position of the ROI 3D box and the ROI 3D box size to adjust the size and position of the ROI 3D box.
（4）Click Save to save the current ROI 3D configuration.
（5）If you need to set multiple ROI 3D type areas, click + to add a new ROI 3D configuration; conversely, click - to delete the current ROI 3D configuration.
Each time a new ROI 3D configuration is added, the system names it by default according to the addition sequence. Custom ROI 3D names are also supported; and the size and position of the 3D box inherit the configuration of the current ROI 3D box. Then set the ROI 3D according to the steps above.
Note
Under the same ROI configuration, when multiple ROI 3D items are set, the ROI 3D types cannot be duplicated. For example, ROI3D_1 and ROI3D_2 cannot both be set to Workspace.
2.4 Delete ROI
To delete the current ROI configuration, click Delete on the ROI configuration page, or right-click the ROI configuration to be deleted in the ROI list and click Delete.
3. Select ROI
When configuring task information, you need to select the required ROI configuration. A task can select only one ROI configuration, and multiple tasks can use the same ROI configuration.
