Hi team, background_subtraction_demo_gapi failed with person-bicycle-car-detection.mp4 from Media Files: https://storage.openvinotoolkit.org/data/test_data/videos.

How to download:
wget https://storage.openvinotoolkit.org/data/test_data/videos/person-bicycle-car-detection.mp4

[Environment]
OS: Ubuntu 22.04.2 LTS
Kernel: 5.15.0-86-generic
Openvino: 2023.3.0
OpenCV: 4.9.0

[Log]
$ ./background_subtraction_demo_gapi -m $HOME/workshop/kazuki/open_model_zoo/demos/background_subtraction_demo/cpp_gapi/intel/instance-segmentation-security-0002/FP32/instance-segmentation-security-0002.xml -i person-bicycle-car-detection.mp4 -at maskrcnn
[ INFO ] OpenVINO
[ INFO ] version: 2023.3.0
[ INFO ] build: 2023.3.0-13775-ceeafaf64f3
[ INFO ] The background matting model /home/yuzhang3/workshop/kazuki/open_model_zoo/demos/background_subtraction_demo/cpp_gapi/intel/instance-segmentation-security-0002/FP32/instance-segmentation-security-0002.xml is loaded to CPU device.
[ ERROR ] [ GENERAL_ERROR ] Exception from src/plugins/intel_cpu/src/infer_request.cpp:393:
Can't set the input tensor with name: image, because the model input (shape=[1,3,?,?]) and the tensor (shape=(1.432.768.3)) are incompatible