Howard Mahé

Hi TL;DR: just use a ResNet-101 trained on ImageNet without dilated convolutions. In my opinion, if one consider a ResNet-101 with dilated convolution as described by Yu and Kolten [1]...

I don't know, only the author could answer. anyway, the main constribution of PSPNet is (1) the PSP module and (2) large crop, large batch and fine tuning BN parameters...

Regarding the 1/4 low resolution image, you are right, it is not used in the provided prototxt. Instead there is a kind of hack of the 1/4 branch using the...

I forced pip reinstalling hydra-core 1.0.6. Btw, docker/requirements.txt is correct and use hydra-core 1.0.6 but the setup.py incorrectly mentions hydra-core>=1.0

It's a bit hidden in the README.md **ILSVRC** Model A: https://cloudstor.aarnet.edu.au/plus/index.php/s/V7dncO4H0ijzeRj Model A1: https://cloudstor.aarnet.edu.au/plus/index.php/s/NOPhJ247fhVDnZH **Pascal VOC12** Model A1, 2 conv.: https://cloudstor.aarnet.edu.au/plus/index.php/s/YqNptRcboMD44Kd Model A1, 2 conv. (COCO): https://cloudstor.aarnet.edu.au/plus/index.php/s/JKWePbLPlpfRDW4 **Cityscapes** Model A1,...

@zhiweige models download works fine for me and model size shown on the website is correct. You should try using a Download Manager if your connection with the server is...

Hi @pudae, thanks a lot for this well-written code. Regarding the _pspnet_v1_50.ckpt_ model, I don't know if this refers to (1) the pre-trained model on ImageNet or (2) the one...

@Heidisnaps You don't really need .ckpt file. One can just restore variables using the checkpoint. - Create a file `checkpoint` in `/path/to/checkpoint/` directory and add the following lines inside: ```...

I gave a quick look at test_model.py, I think you should better use test_segmentation. There are already in this script some useful lines to restore a checkpoint, like: ``` variables_to_restore...

Thanks a lot for your work. I'm running Stock Extensions v24 for GNOME Shell 42 on Ubuntu 22.04. I managed to tweak the patch to update the `main_GS_42` branch (corresponding...