search

seoungwugoh / STM

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

Question about weight initialisation #10

Closed gaomingqi closed 4 years ago

gaomingqi commented 4 years ago

Hi, thanks for sharing your great work!

I am trying to reproduce the training code and I have 2 questions about how to initialise the weights in STM model: 1. In your released code, the backbone network (ResNet-50) uses the weights pre-trained on ImageNet to extract features from video sequence, so does that mean this network module is not fine-tuned further during training? 2. How to initialise the weights in the Decoder module and the convolutional layers for computing key and value features? With the ones generated randomly or pre-trained on some segmentation datasets?

Thank you so much for your consideration and look forward to hearing from you soon!

seoungwugoh commented 4 years ago

@gaomingqi Hi, here is my answers: 1) it was just for initialization, it is fine-tuned during training. 2) Initialization for weights except for ResNet is done randomly (through Kaiming He initialization). Note that we have two training stages: pre-training on images and main training on videos. There is no pre-pre training.

gaomingqi commented 4 years ago

Thank you so much for your reply!