segmentation_models.pytorch

segmentation_models.pytorch copied to clipboard

More classes result in smaller model?

1

I'm not familiar with the internal structure of the Unet model. However, I have one assumption, which I think is reasonable, that when I increase the number of output channels,...

fynv

How to correct contribute example with pt-lightning from my fork to original repo

2

I used this awesome repo for my own work and added pt-lightning for my work. Now I want to be usefull and add this example to orig repo. Should I...

IamSVP94

epsilon in jaccardscore and dicescore is not read as a torch tensor

2

suggestion - add line: eps=torch.tensor(1e-07, dtype=torch.float32) or eps=torch.tensor(1e-07, dtype=gt.dtype)

TomatoGreen2

examples (notebooks) have not been updated

1

https://github.com/Lightning-AI/pytorch-lightning/pull/16520 error: PetModel.on_validation_epoch_end() missing 1 required positional argument: 'outputs'

ChingisBadmaev

data preproccessing : mask = to_tensor

4

my masks shape is (512, 512), when i start the fitting i get an error of miss matching axes , i think the problem is here : ### **def to_tensor(x,...

raidanis

`model = smp.Unet( encoder_name="vgg19", # choose encoder, e.g. mobilenet_v2 or efficientnet-b7 encoder_weights="imagenet", # use `imagenet` pre-trained weights for encoder initialization in_channels=3, # model input channels (1 for gray-scale images, 3...

chahatsagarmain

The latest version of Pillow does not support Image. LINEAR. A similar option that we can use is Image.BILINEAR. Reff: [PIL.Image.LINEAR no longer exists #5010](https://github.com/facebookresearch/detectron2/issues/5010)

PavanPHV

Bumps [timm](https://github.com/huggingface/pytorch-image-models) from 0.9.7 to 1.0.3. Release notes Sourced from timm's releases. Release v1.0.3 May 14, 2024 Support loading PaliGemma jax weights into SigLIP ViT models with average pooling. Add...

dependabot[bot]

Update timm to 1.* version and support more encoders

3

Make most of the models from timm compatible with decoders (including transformers and convnext models)

qubvel

Bumps [timm](https://github.com/huggingface/pytorch-image-models) from 0.9.7 to 1.0.7. Release notes Sourced from timm's releases. Release v1.0.7 June 12, 2024 MobileNetV4 models and initial set of timm trained weights added: model top1 top1_err...

dependabot[bot]