Box 3. Phases of the training processes (Faster R-CNN training model).

Training processes: Different Phases of Faster R-CNN training model:
Phase 1:	After initializing the RPN structure with the pre-trained framework, the RPN is trained. The model’s distinctive value and RPN are revised when the training is finished
Phase 2:	The Faster R-CNN architecture is formed. Subsequently the proposal is calculated by utilizing the trained RPN and then the proposal is sent to the Faster R-CNN network. Following this, the network is trained. Then the model and the uniqueness of the Faster R-CNN is updated through the training process
Phase 3:	The RPN network is initialized by employing the model that was formed in the Phase 2. Then a second training is carried out on the RPN network. The RPN’s distinctive value is altered at the time of the training procedure while the model parameters remain unchanged.
Phase 4:	The model variables stated in Phase 3 are kept unaltered. The Faster R-CNN architecture is formed and trained the network for the 2nd attempt to optimize the specifications