How to run the inference with temporal consistency module

[myselfbritty]

Dear Authors,
Thank you for sharing your work. While trying to run the code, inference.sh, which argument is corresponding for enabling the temporal consistency module so that I can reproduce the results for (T- tick). I could run the code for 2017 dataset with the following .sh instructions but the results are similar to ISINet without Temporal consistency module and without Data augmentation. Could you please help me understand how to run it with Temporal consistency module?
Also, the file inference.sh does not contain the following code but a different version.

python -W ignore main.py --inference --model FlowNet2 --batch_size batch_size --number_workers num_workers
--inference_dataset RobotsegTrackerDataset \ --inference_dataset_img_dir /path/to/images \ --inference_batch_size batch_size
--inference_dataset_coco_ann_path /path/to/coco/annotations/file.json
--inference_dataset_segm_path /path/to/mask-rcnn/inference/segm.json
--inference_dataset_ann_dir /path/to/annotations
--inference_dataset_cand_dir /path/to/save/candidates \ --inference_dataset_nms 'True'
--save /path/to/save/predictions
--inference_dataset_dataset '2017' or '2018'
--inference_dataset_maskrcnn_inference 'True'
--assignment_strategy 'weighted_mode' \ --inference_dataset_prev_frames 7
--threshold 0.0 for 2017 and 0.5 for 2018
--resume /path/to/flownet/checkpoint --num-classes number_of_classes