Rotated Object Detection

Rotated Object Detection is similar to object detection but can also detect the rotation state of the objects.

After completing the model annotations, refer to the video in the Training section to create dataset versions and train/deploy the model.

Use Case Scenarios

Rotated Object Detection is suitable for determining whether an object appears in a scene, the number of times the object appears, and the object's rotation state.

Object Detection can be used to identify the number of occurrences and approximate locations of one or more objects in a scene. While object detection can identify multiple objects simultaneously, it does not provide precise locations. For precise locations, you may use Instance Segmentation or Keypoint Detection models.

Annotation Methods

If you have a pre-trained model, you can use annotation tools to assist with labeling, then review and correct the annotations

First, use the rectangle annotation tool to mark the bounding box, then move the mouse to rotate the box.

Repeat the annotation for all objects in the scene. If there are no objects in the scene, mark it as empty.

Notes

Rotated Object Detection models only support rectangular annotations, so the annotated areas can slightly overlap but should not be completely overlapping.
Similar to other annotation models, avoid annotating objects that are largely covered by other objects. Choose the most visible or topmost objects for annotation.

Practice

Download practice data to obtain rotated_object.zip

After extracting, you will get 11 images and annotation (.json) files. Please upload only the images to DaoAI World for annotation practice. Afterward, you can upload both images and annotation files to compare the results.