Aller au contenu
ModelPrésentéExterne
Télécharger

LN Segmentation Sweep v2

A unet model for multilabel image segmentation trained with sliding window approach.

Sourcehuggingface.co/aholk/LN_segmentation_sweep_v2/tree/90994c4fb1866ea2ce00de88589c949129151b34
TypeOtherVersionRevision 90994c4CClasse

Source du modèle

Type
Other
Taille du fichier
132 MB
Version
vRevision 90994c4
Mis à jour
8 oct. 2026
Téléchargements
5
Source
Hugging Face

Description de la source

LN_segmentation_sweep_v2

A unet model for multilabel image segmentation trained with sliding window approach.

Model Description

Wandb Parameters

ParameterValue
data_pathGleghornLab/Semi-Automated_LN_Segmentation_10_11_2025
img_size128
downsample_factor1
num_channels3
batch_size16
lr1.7122348637490954e-05
epochs100
patience10
weight_decay8.29726636990404e-05
model_typeunet
n_filts32
t3
k3
augmentFalse
normTrue
keep0.06990272917761037
pruning_factor0.019243240405735405
output_dirpooled_metrics_hev_settings
deviceNone
num_workers4
prefetch_factor2
wandb_projectsegmentation-sweep
wandb_run_namehev-only-repro-pooled
wandb_modeonline
push_to_hubTrue
hub_model_idaholk/LN_segmentation_sweep_v2
skip_reportFalse
sweep_modeFalse
num_params34527236
num_classes4

Model Parameters

ParameterValue
num_channels3
num_classes4
n_filts32
t3
k3
img_size128
normTrue
model_archunet
transformers_version5.9.0
architectures["UNetForSegmentation"]
output_hidden_statesFalse
return_dictTrue
dtypefloat32
chunk_size_feed_forward0
is_encoder_decoderFalse
id2label{"0": "LABEL_0", "1": "LABEL_1"}
label2id{"LABEL_0": 0, "LABEL_1": 1}
problem_typeNone
_name_or_path
batch_size16
downsample_factor1.0
model_typesegmentation
output_attentionsFalse

Performance Metrics

MetricMeanClass 0Class 1Class 2Class 3
Dice0.81690.71880.81960.81810.9112
IoU0.69610.56100.69430.69230.8369
F10.81690.71880.81960.81810.9112
MCC0.81240.72610.81710.81340.8928
ROC AUC0.97680.97260.99230.95350.9888
PR AUC0.88210.80460.89600.86520.9627

Usage

import numpy as np
from model import MODEL_REGISTRY, SegmentationConfig

# Load model
config = SegmentationConfig.from_pretrained("aholk/LN_segmentation_sweep_v2")
model = MODEL_REGISTRY["unet"].from_pretrained("aholk/LN_segmentation_sweep_v2")
model.eval()

# Run inference on a full image with sliding window
image = np.random.rand(2048, 2048, 3).astype(np.float32)  # Your image here
probs = model.predict_full_image(
    image,
    dim=128,
    batch_size=16,
    device="cuda"  # or "cpu"
)
# probs shape: (num_classes, H, W) with values in [0, 1]

# Threshold to get binary masks
masks = (probs > 0.5).astype(np.uint8)

Training Plots

Training Loss Dice Curves IoU Curves MCC Curves Best Validation

Citation

If you use this model, please cite:

@software{windowz_segmentation,
  title={Multilabel Image Segmentation with Sliding Window U-Net},
  author={Gleghorn Lab},
  year={2025},
  url={https://github.com/GleghornLab/ComputerVision2}
}

Sources

1 source
Hugging FacePrincipalVersion 90994c4fb1866ea2ce00de88589c949129151b34
huggingface.co/aholk/LN_segmentation_sweep_v2/tree/90994c4fb1866ea2ce00de88589c949129151b34

Vérifié 25 août

Artefacts du modèle

1 artefact

model.safetensors

safetensors · 132 MB · SHA-256 f1eff854c715…5197 · Hugging Face

Télécharger

Extraits de sources

2 extraits
Readme25 aoûthuggingface.co/aholk/LN_segmentation_sweep_v2

--- license: mit tags: - image-segmentation - multilabel - unet - pytorch - medical-imaging library_name: transformers pipeline_tag: image-segmentation --- # LN_segmentation_sweep_v2 A unet model for multilabel image segmentation trained with sliding window approach. ## Model Description ## Wandb Parameters | Parameter | Value | |-----------|-------| | data_path | GleghornLab/Semi-Automated_LN_Segmentation_10_11_2025 | | img_size | 128 | | downsample_factor | 1 | | num_channels | 3 | | batch_size | 16 | | lr | 1.7122348637490954e-05 | | epochs | 100 | | patience | 10 | | weight_decay | 8.29726636990404e-05 | | model_type | unet | | n_filts | 32 | | t | 3 | | k | 3 | | augment | False | | norm | True | | keep | 0.06990272917761037 | | pruning_factor | 0.019243240405735405 | | output_dir | pooled_metrics_hev_settings | | device | None | | num_workers | 4 | | prefetch_factor | 2 | | wandb_project | segmentation-sweep | | wandb_run_name | hev-only-repro-pooled | | wandb_mode | online | | push_to_hub | True | | hub_model_id | aholk/LN_segmentation_sweep_v2 | | skip_report | False | | sweep_mode | False | | num_params | 34527236 | | num_classes | 4 | ## Model Parameters | Parameter | Value | |-----------|-------| | num_channels | 3 | | num_classes | 4 | | n_filts | 32 | | t | 3 | | k | 3 | | img_size | 128 | | norm | True | | model_arch | unet | | transformers_version | 5.9.0 | | architectures | ["UNetForSegmentation"] | | output_hidden_states | False | | return_dict | True | | dtype | float32 | | chunk_size_feed_forward | 0 | | is_encoder_decoder | False | | id2label | {"0": "LABEL_0", "1": "LABEL_1"} | | label2id | {"LABEL_0": 0, "LABEL_1": 1} | | problem_type | None | | _name_or_path | | | batch_size | 16 | | downsample_factor | 1.0 | | model_type | segmentation | | output_attentions | False | ## Performance Metrics | Metric | Mean | Class 0 | Class 1 | Class 2 | Class 3 | |--------|------|--------|--------|--------|--------| | Dice | 0.8169 | 0.7188 | 0.8196 | 0.8181 | 0.9112 | | IoU | 0.6961 | 0.5610 | 0.6943 | 0.6923 | 0.8369 | | F1 | 0.8169 | 0.7188 | 0.8196 | 0.8181 | 0.9112 | | MCC | 0.8124 | 0.7261 | 0.8171 | 0.8134 | 0.8928 | | ROC AUC | 0.9768 | 0.9726 | 0.9923 | 0.9535 | 0.9888 | | PR AUC | 0.8821 | 0.8046 | 0.8960 | 0.8652 | 0.9627 | ## Usage ```python import numpy as np from model import MODEL_REGISTRY, SegmentationConfig # Load...

Note25 aoûthuggingface.co/aholk/LN_segmentation_sweep_v2

Source context: 5 downloads · 0 likes · Pipeline image-segmentation · Library transformers · Repo aholk/LN_segmentation_sweep_v2