Developing a keypoint detection application
Besides detecting objects, keypoint detection also indicates important parts of the detected objects called keypoints. These keypoints describe the detected object’s essential trait. This trait is often invariant to image rotation, shrinkage, translation, or distortion. The following sections detail the steps to develop a keypoint detection application using Detectron2 pre-trained models.
Selecting a configuration file
Detectron2 also provides a list of cutting-edge algorithms pre-trained for keypoint detection for human objects. For instance, Figure 2.7 shows the list of Mask R-CNN models pre-trained on the COCO Person Keypoint Detection dataset.
Figure 2.7: COCO Person Keypoint Detection baselines with Keypoint R-CNN
In this specific case, we select the X101-FPN pre-trained model. Again, the link to the configuration file is linked with the model name in the first column, and we only use the part...
