search

seoungwugoh / STM

Video Object Segmentation using Space-Time Memory Networks
405 stars 81 forks source link

Test input size #25

Open moonlight99 opened 4 years ago

moonlight99 commented 4 years ago

Thanks for the good work done. I tried to repeat your results. As it is written in your wonderful article, I trained at a size of 384x384. My trained model works at 384x384 input size. But when I try to input the full size from the Davis dataset (as you have in the demo script), I get trash drawn. Have you encountered such a problem and how did you solve it? Waiting for your reply!

seoungwugoh commented 4 years ago

@moonlight99 Did you make any changes in model.py? It happens when the model is not fully convolutional. The provided weights is also trained only with 384x384 patches but there was no problem with testing on other resolutions.