Takes synchronised sensor data as input and outputs processed visual data.
- Outputs detected object labels.
- Outputs positions and orientations for each detected object
- Generates 3D reconstruction.
TODO Next
- Initiate Dense 3D model with Elastic Fusion
- For each new frame, write code for object detection (both in 2D and 3D)
- Set-up transformations that link each Camera pose with object poses i.e. track object and camera poses over time
TODO Later -Propogate object detections both backward and forwards in time. -(Object level Kalman Filtering)