🖥
🖥
🖥
🖥
AugeLab Studio Manual
English
Ask or search…
⌃K
Comment on page

Object Detection - Custom

Object Detection - Custom Node Documentation

The Object Detection - Custom node in AugeLab Studio is used to detect objects of specific classes in an input image using a custom object detection model. It allows you to specify the weight file, config file, and class file for the object detection model, and provides options to select the desired object classes to detect.

Node Overview

The Object Detection - Custom node takes an image as input and applies a custom object detection model to identify and localize objects of specific classes. It provides the following outputs:
  • Processed Image: The input image with the object detection results visualized.
  • Object Count: The total number of objects detected.
  • Object Locations: The coordinates of the bounding boxes around the detected objects.
  • Object Sizes (w, h): The width and height of the bounding boxes around the detected objects.
  • Object Class: The class labels of the detected objects.

Inputs

The Object Detection - Custom node requires the following input:
  • Image: The input image to be processed.

Outputs

The Object Detection - Custom node provides the following outputs:
  • Processed Image: The input image with the object detection results visualized.
  • Object Count: The total number of objects detected.
  • Object Locations: The coordinates of the bounding boxes around the detected objects.
  • Object Sizes (w, h): The width and height of the bounding boxes around the detected objects.
  • Object Class: The class labels of the detected objects.

Node Interaction

  1. 1.
    Drag and drop the Object Detection - Custom node from the node library onto the canvas in AugeLab Studio.
  2. 2.
    Connect the input image to the Image input socket of the Object Detection - Custom node.
  3. 3.
    Click the "Open Weight File" button and select the weight file for the custom object detection model.
  4. 4.
    Click the "Open Config File" button and select the config file for the custom object detection model.
  5. 5.
    Click the "Open Class File" button and select the class file for the custom object detection model.
  6. 6.
    The table view displays the available class names. Check the checkboxes next to the desired classes to select them for detection.
  7. 7.
    Configure the confidence threshold using the slider.
  8. 8.
    The output sockets provide the following information:
    • Processed Image: The input image with the object detection results visualized.
    • Object Count: The total number of objects detected.
    • Object Locations: The coordinates of the bounding boxes around the detected objects.
    • Object Sizes (w, h): The width and height of the bounding boxes around the detected objects.
    • Object Class: The class labels of the detected objects.
  9. 9.
    Connect the output sockets to other nodes for further processing or analysis.

Implementation Details

The Object Detection - Custom node is implemented as a subclass of the CalcNode base class. It overrides the evalImplementation method to perform the object detection.
  • The node reads the input image from the input socket.
  • The custom object detection model is initialized using the provided weight file, config file, and class file.
  • The selected object classes for detection are obtained from the table view.
  • The object detection model is applied to the input image, and the results are obtained.
  • The processed image with the object detection results, along with the count, locations, sizes, and class labels of the detected objects, is returned.

Usage

  1. 1.
    Drag and drop the Object Detection - Custom node from the node library onto the canvas in AugeLab Studio.
  2. 2.
    Connect the input image to the Image input socket of the Object Detection - Custom node.
  3. 3.
    Click the "Open Weight File" button and select the weight file for the custom object detection model.
  4. 4.
    Click the "Open Config File" button and select the config file for the custom object detection model.
  5. 5.
    Click the "Open Class File" button and select the class file for the custom object detection model.
  6. 6.
    Check the checkboxes next to the desired classes in the table view to select them for detection.
  7. 7.
    Configure the confidence threshold using the slider.
  8. 8.
    The output sockets provide the following information:
    • Processed Image: View the input image with the object detection results visualized.
    • Object Count: Get the total number of objects detected.
    • Object Locations: Get the coordinates of the bounding boxes around the detected objects.
    • Object Sizes (w, h): Get the width and height of the bounding boxes around the detected objects.
    • Object Class: Get the class labels of the detected objects.
  9. 9.
    Connect the output sockets to other nodes for further processing or analysis.
  10. 10.
    Continue building your pipeline by connecting other nodes as needed.

Notes

  • The Object Detection - Custom node applies a custom object detection model to identify and localize objects in an image.
  • The weight file, config file, and class file for the custom object detection model must be provided.
  • The node allows you to select the desired object classes for detection using the table view.
  • The confidence threshold can be adjusted to control the sensitivity of the object detection. Higher values result in stricter detection.
  • The Object Detection - Custom node provides both the processed image with the object detection results and additional information about the detected objects.
  • Use the outputs of the Object Detection - Custom node to perform further analysis or visualization, or as input for other nodes in the pipeline.
  • Consider combining the Object Detection - Custom node with other nodes to create a more comprehensive AI application or analysis pipeline.
Function Blocks - AI Applications - Previous
Object Detection
Next - Function Blocks - AI Applications
Object Detection Train
Last modified 4mo ago