Proposed approach

                  - Geometric calibration
                       - camera, camera to target (using target model)
                       - illuminators to target (using cast shadows)

                  - Compile training data set from known targets

                  - Validate on new views, new targets