Skip to content
Other

DeepContrail segmentation v0.1.0

сегментация инверсионных следов на снимках GOES-16

2 min read

Модель от EUROCONTROL (eurocontrol-asu) для задачи image-segmentation: по 3-канальному false-colour композиту GOES-16 ABI она выдаёт для каждого пикселя вероятность инверсионного следа. Архитектура — UPerNet с энкодером CAFormer-B36 (timm caformer_b36.sail_in22k_ft_in1k_384), 102,6 млн параметров; обучена на OpenContrails (экспорт 20250305).

Ключевые особенности

  • Вход — стандартный ash RGB: R = band15−14, G = band14−11, B = band14; сцена 256×256 масштабируется до 512 и нормализуется.
  • Выход — карта вероятностей в [0,1] в исходном разрешении; порог по умолчанию 0,5.
  • Рекомендуется TTA: усреднение 8 отражений и поворотов.
  • Обучение на soft-labels — доля аннотаторов, отметивших след; 18 079 сцен из 20 088, 2 009 в валидации.
  • Веса под Apache 2.0, код под EUPL-1.2; есть отдельный instance-пайплайн, но он не входит в v0.1.0.

Пример

import numpy as np
from deepcontrail.core import false_color_from_spectral, read_band_manifest
from deepcontrail.segmentation import load_pretrained, predict
model = load_pretrained('eurocontrol-asu/deepcontrail-segmentation', revision='v0.1.0')
bands = np.load(f'{root}/spectral_bands/test/{record_id}.npy')
composite = false_color_from_spectral(bands, model.config.false_color, read_band_manifest(root))
probabilities = predict(model, composite, tta=model.config.tta)
mask = probabilities > model.config.threshold
ПодробнееGlobal Dice на тесте: 0.6960 с TTA и порогом 0.5 (single pass — 0.6892).
Обучение: 40 эпох, batch 32, AdamW, BCE + 0.2 × Dice, bf16; 18 079 сцен.
Ограничения: только GOES-16 ABI 256×256 (~2 км/пиксель), другие размеры и неквадратные сцены пока не поддерживаются.
Для старых 20230419/Kaggle-масок нужен pixel_shift=0.5; instance-пайплайн не входит в v0.1.0.
Discuss Telegram 1 53

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # eurocontrol-asu/deepcontrail-segmentation
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download eurocontrol-asu/deepcontrail-segmentation --local-dir ~/models/deepcontrail-segmentation
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of deepcontrail-segmentation (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

eurocontrol-asu/deepcontrail-segmentation
Author
eurocontrol-asu
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

531

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments