lucid-dl/fcn-resnet-101

FCN ResNet-101

Long et al., 2015 — Fully Convolutional Networks for Semantic Segmentation (arXiv:1411.4038)

Lucid port of torchvision/FCN_ResNet101_Weights.COCO_WITH_VOC_LABELS_V1, converted to Lucid-native safetensors.

Available weights

Tag	miou	pixel_acc	Params	GFLOPs	Size	Source
COCO_WITH_VOC_LABELS_V1 (default)	63.7	91.9	54.3M	232.738	207.67 MB	torchvision

Usage

import lucid.models as models
from lucid.models.weights import FCNResNet101Weights

# default tag
model = models.fcn_resnet101(pretrained=True)

# explicit tag (enum or string)
model = models.fcn_resnet101(weights=FCNResNet101Weights.COCO_WITH_VOC_LABELS_V1)
model = models.fcn_resnet101(pretrained="COCO_WITH_VOC_LABELS_V1")

# preprocessing travels with the weights
weights = FCNResNet101Weights.COCO_WITH_VOC_LABELS_V1
preprocess = weights.transforms()
out = model(preprocess(image)[None])
# SemanticSegmentationOutput: per-pixel class logits (B, C, H, W)
seg = out.logits.argmax(axis=1)  # (B, H, W) class indices

Conversion

Converted from torchvision/FCN_ResNet101_Weights.COCO_WITH_VOC_LABELS_V1 via python -m tools.convert_weights fcn_resnet101 --tag COCO_WITH_VOC_LABELS_V1. Key mapping + numerical parity verified against the source.

License

bsd-3-clause — inherited from the original weights.

Citation

@inproceedings{long2015fully,
  title={Fully Convolutional Networks for Semantic Segmentation},
  author={Long, Jonathan and Shelhamer, Evan and Darrell, Trevor},
  booktitle={CVPR}, year={2015}
}