Model cards: volume
Collection
fastforks model cards, task volume. • 3 items • Updated
TotalSegmentator (fastforks)
TotalSegmentator (Wasserthal et al., Radiology: AI 2023): 117 anatomical structures (organs, bones, muscles, vessels) in CT, from nnU-Net v2 3D U-Nets. total merges five 1.5 mm part networks, total-3mm is the single fast 3 mm network. Gives one integer label volume on the input grid plus the label names.
This repo is the model card only: it holds no weights. fastforks downloads the authors' weights from the links below and checks them against the SHA-256 in the model card. Code: https://github.com/kleinicke/fastforks. Kernels for the kernels backend: https://huggingface.co/fastforks/kernels. Original code: https://github.com/wasserth/TotalSegmentator at 184fa4765456.
Task: volume. Tier: experimental. Outputs: labels. Backends: portable, kernels.
Input: a CT volume [D, H, W] in Hounsfield units and its voxel spacing in mm (volume.read_nifti reads both from a NIfTI file). Output: a class index per voxel [D, H, W] on the input grid (0 = background). Encoder models (nnFoundation) take a CT or MRI volume and give its encoder feature maps.
Variants
Weights are the authors' files (not hosted here). SHA-256 shortened to 12 hex digits; the full hashes are in card.toml.
Licence
Original-code strings, recorded as they are (research-use package; check them before any other use). Code: Apache-2.0. Licence files: LICENSE.
Training data: TotalSegmentator dataset v2 (1228 CT scans, 117 structures) (1559 CT scans of the TotalSegmentator dataset v2 (CC BY 4.0). Only the weights licence counts for the commercial flag.).
Licence evidence: https://github.com/wasserth/TotalSegmentator/blob/184fa4765456492b2ab81a981cc387465906b003/LICENSE ; https://github.com/wasserth/TotalSegmentator/blob/184fa4765456492b2ab81a981cc387465906b003/README.md
Input size
any D x H x W volume and its voxel spacing (resampled to the model spacing, sliding window)
Tested up to (portable, fp16, RTX 5060 Ti 16 GB): totalsegmentator-total 512x512x512 voxels (5.5 GiB peak allocated, 5.8 reserved); totalsegmentator-total-3mm 512x512x512 voxels (4.3 GiB peak allocated, 5.1 reserved)
Speed and memory
Volume, synthetic
| model | input | time (ms): original fp32 | original | portable | kernels | kernels speed-up: vs original | vs portable | peak memory (MiB, allocated / reserved): original | portable | kernels |
|---|---|---|---|---|---|---|---|---|---|---|
totalsegmentator-total |
200x200 | 47779 | 45630 | 1927 | 1740 | 26.2x | 1.1x | 2640 / 2874 | 2605 / 2944 | 2349 / 2704 |
totalsegmentator-total-3mm |
200x200 | 7304 | 7270 | 67.1 | 63.5 | 114.4x | 1.1x | 2010 / 2684 | 1092 / 1508 | 1037 / 1410 |
Startup and first call
| model | load (s): original | portable | kernels | first call (s): original | portable | kernels |
|---|---|---|---|---|---|---|
totalsegmentator-total |
0.07 | 0.77 | 0.77 | 54.51 | 2.10 | 1.88 |
totalsegmentator-total-3mm |
0.06 | 1.18 | 1.19 | 12.48 | 0.26 | 0.21 |
torch.compile and TensorRT
| model | input | time: portable | compile | TensorRT fp16 | kernels | preparation: compile (cold first call) | ONNX + engine build | peak MiB (alloc / reserved): portable | compile | kernels | TensorRT (engine + torch) |
|---|---|---|---|---|---|---|---|---|---|---|---|
totalsegmentator-total |
200x200 | 1927 ms | fails: no exportable module (multi-network sliding-window pipeline) | fails: no exportable module (multi-network sliding-window pipeline; | 1740 ms | – | – | 2605 / 2944 | – | 2349 / 2704 | – |
totalsegmentator-total-3mm |
200x200 | 67.1 ms | fails: no exportable module (multi-network sliding-window pipeline) | fails: no exportable module (multi-network sliding-window pipeline; | 63.5 ms | – | – | 1092 / 1508 | – | 1037 / 1410 | – |
Fixed-size cells: time (ms) / peak allocated / peak reserved memory (MiB).
Hardware, protocol and caveats: benchmarks (RTX 5060 Ti 16 GB, sm_120). Original = the original code named above at the precision of portable (fp16, or (fp32) where fp16 is unsupported or not finite); original fp32 = its default fp32 run. Peak memory is the torch allocator's.
Accuracy
On public datasets
A small subset of real CT cases of the TotalSegmentator dataset (not the full test split), each run at its card's spacing (1.5 or 3 mm) in the same pipeline; mean over the cases of the Dice (0 to 1) averaged over the classes present, against the ground-truth segmentations (vertebrae_L6 of the subset has no counterpart and is skipped). Original = TotalSegmentator 2.18.0, default fp16 autocast. Original = the original code named under Details, same checkpoint and protocol.
| variant | dataset | original fp16 | portable | kernels | note |
|---|---|---|---|---|---|
totalsegmentator-total |
TotalSegmentator test subset (7 CT cases), 1.5 mm, mean Dice | 0.9365 | 0.9365 | 0.9365 | |
totalsegmentator-total-3mm |
TotalSegmentator test subset (8 CT cases), 3 mm, mean Dice | 0.8285 | 0.8285 | 0.8285 |
Measured on NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 2d71f180, 2026-11-14.
Parity with the original code
Every variant is checked against reference outputs of the original code at its pinned commit. CPU, float32: max |diff| <= 0.0 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.003 and p99 <= 0.5 (absolute error of the output labels).
Usage
from fastforks import volume
model = volume.load("totalsegmentator-total") # mode="auto": the preferred backend from recorded measurements
result = model(ct, spacing=spacing) # ct: [D, H, W] float32 HU volume; spacing: (z, y, x) in mm
result.labels # [D, H, W] uint8 label per voxel; result.class_names[k] names label k
result.backend # what ran, e.g. Backend(mode='portable', device='cuda:0', dtype='float16')
Install with pip install fastforks (see https://github.com/kleinicke/fastforks). Other variants: totalsegmentator-total, totalsegmentator-total-3mm.
Kernels backend
volume.load("totalsegmentator-total", mode="kernels") runs our CUDA kernels (vol_norm). It needs a prebuilt kernel for your GPU architecture (downloaded from https://huggingface.co/fastforks/kernels on first use) and a recent NVIDIA driver; if there is none it raises BackendUnavailableError with the reason and never falls back silently. mode="auto" picks kernels when it can run and logs the choice.