pytorch-image-models icon indicating copy to clipboard operation
pytorch-image-models copied to clipboard

The largest collection of PyTorch image encoders / backbones. Including train, eval, inference, export scripts, and pretrained weights -- ResNet, ResNeXT, EfficientNet, NFNet, Vision Transformer (ViT)...

Results 237 pytorch-image-models issues
Sort by recently updated
recently updated
newest added

# **Motivation** When using `timm` as a backbone library, users often switch between different model architectures, such as CNNs and Vision Transformers. These models frequently have different `__init__` signatures. For...

## Fix infinite recursion risk in ImageDataset error handling ### Changes - Added maximum retry limit to prevent infinite recursion when loading corrupted/missing files - Improved error messages and logging...

I use validate.py to validate the performance of the model on a new dataset, but I need these as indicators. Is it possible to increase these indicators?

enhancement

Given the popularity of Agglomerative Vision Foundation Models such as RADIO, RADIOv2.5. It would be great if they can be added into TIMM. https://github.com/NVlabs/RADIO

enhancement

Refactor the DPN model in `timm` style with support for `forward_intermediates`, and map the checkpoint weights to ensure output is the same.

**Is your feature request related to a problem? Please describe.** I rely on the `forward_intermediates()` API for object detection models, and I'm experimenting with ViT-g and would like to try...

enhancement
help wanted

Hi, timm is excellent and helps me a lot for my research. Thanks for your great work! Do you have the plan to support more strong vision models like Depth...

enhancement

While playing around with Flux-Redux, which used siglip-so400m, I noticed that loading time is pretty slow. I think we can bring some of the loading-time optimizations for LLMs to timm...

Main train.py does not properly instantiate ImageFolder dataset if --train-split --val-split and --num-classes are specified. It looks like this may only be happening if the main data-dir had additional folders...