Mask R-CNN Example does not work

I want to work with Mask R-CNN according to https://de.mathworks.com/help/vision/ug/example-InstanceSegmentationUsingMaskRCNNDeepLearningExample.html.

If I run the code on my computer (M2021a and b) as well as the "Try this example"-Button on the page, I get the following issue in line 4:

pretrained = load(fullfile(dataFolder,"maskrcnn_object_person_car_v2.mat"));

Unable to find file or directory 'C:\Users\thoma\AppData\Local\Temp\coco\maskrcnn_object_person_car_v2.mat'.

I looked into the directory and it seems to be a typo in either the example or the .mat file on the server: the load function downloaded

maskrcnn_object_person_car.mat

If I try to load it, I get:

Warning: Variable 'net' originally saved as a maskrcnn cannot be instantiated as an object and will be read in as a

uint32.

Edit: I was not aware that I am able to update to 2022a (I have a student license). I will retry it with 2022a and provide an update.

Answer 1

[Thomas Cimiega]

Okay, I managed to install and try it with M2022a and it seems to work. Nevertheless the problem still exists for M2021x users.

[tao wang]

Excuse me, how can I replace it with my own dataset for training?