You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I have a person detection network trained on 64x128 dataset, I used the nonsquare convolution functions but the pooling functions crash because they don't support nonsquare input. I would like to request nonsquare max/ave pooling functions to work with the nonsquare convolution functions. I implemented a slow (non-dsp version) nonsquare pooling, but a dsp version will be much appreciated. Thanks!
The text was updated successfully, but these errors were encountered:
As far as I can tell, the pooling function signature should be adjusted to include both x and y dimensions for pool, stride, pad, etc, and the code slightly adjusted such that dim_im_in should be changed to dim_im_in_x or dim_im_in_y as needed for example.
Same here. I am looking for 1D pooling, but finally, I have to make my own changes.
All I have done is copy and paste the original one and delete all the loop with the y-axis.
I believe you could modify it easily since the x and y are performed separately if you see the comments in the source
Hi, I have a person detection network trained on 64x128 dataset, I used the nonsquare convolution functions but the pooling functions crash because they don't support nonsquare input. I would like to request nonsquare max/ave pooling functions to work with the nonsquare convolution functions. I implemented a slow (non-dsp version) nonsquare pooling, but a dsp version will be much appreciated. Thanks!
The text was updated successfully, but these errors were encountered: