SK-ResNet-34

Li et al., 2019 — Selective Kernel Networks (arXiv:1903.06586)

Lucid port of timm/skresnet34.ra_in1k, converted to Lucid-native safetensors.

Available weights

Tag	acc@1	acc@5	Params	GFLOPs	Size	Source
RA_IN1K (default)	76.956	—	22.3M	—	85.14 MB	timm

Usage

import lucid.models as models
from lucid.models.weights import SKResNet34Weights

# default tag
model = models.sk_resnet_34_cls(pretrained=True)

# explicit tag (enum or string)
model = models.sk_resnet_34_cls(weights=SKResNet34Weights.RA_IN1K)
model = models.sk_resnet_34_cls(pretrained="RA_IN1K")

# preprocessing travels with the weights
weights = SKResNet34Weights.RA_IN1K
preprocess = weights.transforms()
logits = model(preprocess(image)[None]).logits

Conversion

Converted from timm/skresnet34.ra_in1k via python -m tools.convert_weights sk_resnet_34 --tag RA_IN1K. Key mapping + numerical parity verified against the source.

License

apache-2.0 — inherited from the original weights.

Citation

@inproceedings{li2019selective,
  title={Selective Kernel Networks},
  author={Li, Xiang and Wang, Wenhai and Hu, Xiaolin and Yang, Jian},
  booktitle={CVPR}, year={2019}
}