Error reporting out of memory after increasing image size

[FancyFai]

Dear author,
I am a student from hunan university. Thank you very much for the code you provided.However, I encountered a problem in the use, when I increased the image size from 128128 to 256256, there will be a out of memory error . I would like to ask what I need to do if I want to increase the size of the picture.
I am looking forward to your reply.wish you a happy life

[Guo-Xiaoqing]

The main problem exists in the calculation of AAM.

[FancyFai]

The main problem exists in the calculation of AAM.

Thank you very much for your reply. May I ask what modifications I need to make in AAM to make the image larger and still trainable

[Guo-Xiaoqing]

You can try to modify the deformable convolution, which requires a lot of computing resources. In fact, this is also the limitation of AAM. So far, I don't know how to modify matrix operations in deformable convolution. Maybe you can refer to other literatures related to the modification of deformable convolution. FancyFai <notifications@github.com>于2020年5月1日 周五07:17写道：

…

The main problem exists in the calculation of AAM. Thank you very much for your reply. May I ask what modifications I need to make in AAM to make the image larger and still trainable — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIKMY2TJK6XDJJW43FKXA2LRPIBJ3ANCNFSM4MJI3FPA> .

[FancyFai]

You can try to modify the deformable convolution, which requires a lot of computing resources. In fact, this is also the limitation of AAM. So far, I don't know how to modify matrix operations in deformable convolution. Maybe you can refer to other literatures related to the modification of deformable convolution. FancyFai notifications@github.com于2020年5月1日 周五07:17写道：
…
The main problem exists in the calculation of AAM. Thank you very much for your reply. May I ask what modifications I need to make in AAM to make the image larger and still trainable — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIKMY2TJK6XDJJW43FKXA2LRPIBJ3ANCNFSM4MJI3FPA .

Ok, thank you very much for your suggestion. I will look for some literature on deformable convolution and see if I can find a solution.