Detection module overview
ViSP contains a detection module that provides wrapper over optional third-party libraries to enable ViSP faces, barcodes and object detection capabilities.
The source code documentation shows which are the corresponding classes part of the detection module.
Following videos give an overview of this module:
|Face detection thanks to ViSP wrapper over OpenCV haar cascade classifier. The faces are sorted from largest to the smallest in the image. The center of gravity of the largest face can be used to gaze a robot on a human. A tutorial is available here.
|QR code detection thanks to ViSP wrapper over zbar library. The pose of the AR code can then be used to initialize the markerless model-based tracker. A tutorial is available here.
|This video shows how we learn a cube from 3 images and then detect and localize in 3D the cube with respect to the camera. A typical use case consists in using this object detection capability to initialize the markerless model-based tracker. A tutorial is available here.
|The object detection algorithm is based on keypoint matching. This video shows how internally the software is able to match keypoints from one of the 3 learned images with the current image in the bottom-right corner. The matched keypoints are then used to compute the pose of the object in a robust scheme.