how to resize my dataset segmentation including image datastore and pixellabeldastore

while training resnet50 on semantic segmentation Dataset i have this errot

The training images are of size 183×275×3 but the input layer expects images of size

640×640×3.

i want to resize my whole dataset (incluting both datastore)to 640x640x3

Answer 1

[VINAYAK LUHA]

Hi Ahmad.

It is my understanding that the input layer of the resnet50 model expects data of dimensions 640x640x3, but your training images are of dimensions 183x275x3, hence you wish to know a workaround to resize your entire datastores to compatible format.

Refer to the following documentation on how to perform resize image and pixel label as part of semantic segmentation workflows https://www.mathworks.com/help/deeplearning/ug/augment-pixel-labels-for-semantic-segmentation.html#AugmentPixelLabelsForSemanticSegmentationExample-1

Regards,

Vinayak Luha