Albert Mohwald

When I output with any model, I got probabilities per class on each pixel. E.g. when I run ```python model = encoding.models.get_model('DeepLab_ResNeSt101_ADE', pretrained=True).eval() img = torch.randn(1,3,256,256) output = model.evaluate(img) print(output.shape)...

Conda Installation without root

1

Following the [Installation with Conda](https://github.com/NVlabs/imaginaire/blob/c6f74845c699c58975fd12b778c375b72eb00e8d/INSTALL.md) guide, should not I run ``` conda activate imaginaire ``` before executing individual cython installations of third-party libraries? I am asking, because I do not...

Support PyTorch pipeline

2

Related to the #107, in the [fAIr-utilities](https://github.com/hotosm/fAIr-utilities) I would first implement PyTorch pipeline. Then, evaluation of the current SotA models, like cnn-based yolo, mobile-net and smallest ViT-based segment anything would...

Since `Pillow>=10.0.0`, the `Image.ANTIALIAS` resampling is no longer supported and it is changed for `Image.LANCZOS`, see [pillow changelog](https://pillow.readthedocs.io/en/stable/releasenotes/10.0.0.html).