cvlab-unibo / inr2vec Goto Github PK

thanks for sharing.
does this ever tested on medical images, such as ultrasound images ?

Hello,

I am following your code for shape generation and applying to my dataset.
I have already created a dataset in the same format used in your experiments, however, while training INR2vec, I always get OOM during validation step. My RAM and GPU memory keep on increasing during this stage, while during the training iteration it is well below 40% (12 G Titan X), it seems like there is a memory leak or something I am not able to figure out.
I tried reducing the num of query samples, network and embedding size, but still no help, it gets OOM for each sample during validation step.

Could you please help!