Chap12: "spatial_size" parameter --> "canvas_size" from torchvision 0.16

m.taniguchi

(This post is just for your reference.)

In Chap12, some code snippets using "spatial_size" parameter of "BoundingBoxes subclass" get error of "TypeError". (My local environment uses torchvision 0.18.0,Colab uses 0.17.1+cu121 as of today)

According to release note, the parameter has changed its name to "canvas_size" from v0.16.0.

[transforms.v2] Renaming: BoundingBoxes.spatial_size -> BoundingBoxes.canvas_size (#7734)

This worked fine for me.

dvgodoy

Hi @m.taniguchi

Thank you for pointing this out. While the notebooks on GitHub are already updated to reflect this change from "spatial_size" to "canvas_size," we still need to update the code snippets accordingly.

fcioanca

Hi @m.taniguchi

The code snippets have been updated.

Thank you,
Flavia
Linux Foundation Training Team