How to test maskrcnn. Seems like the exported onnx input size is static while the test input size is not fixed

[siahuat0727]

The way I export maskrcnn onnx model

$ python mmdeploy/tools/torch2onnx.py mmdeploy/configs/mmdet/detection/detection_onnxruntime_static.py \
 mmdetection/configs/mask_rcnn/mask-rcnn_r50_fpn_1x_coco.py \
./mask_rcnn_r50_fpn_1x_coco_20200205-d4b0c5d6.pth \
mmdetection/demo/demo.jpg

The way I test maskrcnn onnx model

$ python mmdeploy/tools/test.py mmdeploy/configs/mmdet/detection/detection_onnxruntime_static.py \
mmdetection/configs/mask_rcnn/mask-rcnn_r50_fpn_1x_coco.py \
--model work-dir/end2end.onnx

The error message

RuntimeError: Error in execution: Got invalid dimensions for input: input for the following indices
 index: 3 Got: 1248 Expected: 1216

Environment

$ pip freeze|grep mm
mmcv==2.0.1
-e git+https://github.com/open-mmlab/mmdeploy.git@71a1e6cc1859d8692cea134946bdc249ad0e5400#egg=mmdeploy
mmdeploy-runtime==1.2.0
-e git+https://github.com/open-mmlab/mmdetection@f78af7785ada87f1ced75a2313746e4ba3149760#egg=mmdet
mmengine==0.8.4

I wonder if it is possible to test maskrcnn using onnxruntime with a fixed input size? I would appreciate any feedback. Thank you.