How to run multi-images inference of tensorRT RTMDet instance segmentation?

[miaowu99]

in https://github.com/open-mmlab/mmdetection/tree/main/configs/rtmdet, I only found inference SDK of detection task, how can I run an instance segmentation inference of RTMdet like "from mmdeploy_python import Detector"? or, for inference with downloaded model directly, when I run inference with "from mmdet.apis import DetInferencer", I found that the inference is slow whatever model I choosed...about only 4fps, Why?