how to inference with my own trained model not "demo.py"

Dear,
how to  use my own model for inference, not the "demo.py":

I try this, but it is wrong:
python train.py --config-file ../configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml 
  --input input1.jpg  --output outpath.png  --opts MODEL.WEIGHTS my_model_final.pth

can you tell me the command about how to inference with my own trained model?
Thanks
