To use this library:
- Create a Frame_Handler object
- Set the camera callback to send the frame to Frame_Handler.update_from_camera(frame)
- Call Frame_Handler.process_frame() at whatever frequency desired. Parameters for temporal consistency will need to be optimized for any given run rate.
- Cropped images to send for classification can be found under Frame_Handler.<object type>
- (x, y, z) coordinates for detected objects can be found under Frame_Handler.<object type>_loc
Updates to come:
- Integration with classification models
- In house Haar Cascade models for detection of competition specific objects