Object Detection

Attention

This Step is already out of support and will be excluded from later Mech-Vision versions. Please use the latest Steps Deep Learning Model Package Inference instead.

Function

This Step uses deep learning methods to detect target objects in the scene, and outputs the masks and corresponding labels of the bounding boxes of the target objects.

Usage Scenario

This Step helps use a model from Mech-DLK to make the inference. The results of inference are closely related to the data on which the model was trained.

Input and Output

  • Input:

  1. Color image after ROI setting and scaling.

  2. Depth map corresponding to color image after ROI setting and scaling.

  • Output:

  1. Visualization color image with bounding boxes of the target objects (if the corresponding parameters are set).

  2. Classification labels of detected target objects.

  3. Bounding box masks of detected target objects.

  4. Confidences corresponding to detected target objects.