Skip to content
Other

FarmSAM

компактные student-модели для трекинга животных на ферме

2 min read

FarmSAM — набор student-чекпоинтов для видеотрекера SAM 3. Компактный TinyViT-21M-512 заменяет backbone SAM 3 и дистиллируется на неразмеченных кадрах одной фермы; остальной трекер остаётся без изменений. В карточке: три student-модели для телят, свиней PigTrack и Edinburgh, по 40,66 млн параметров в backbone и 60,20 млн всего вместе с трекером.

Ключевые особенности

  • Backbone на 91% меньше и в 3,6 раза быстрее, чем у SAM 3.
  • Вход 1024×1024, выход карты признаков [1, 1024, 72, 72].
  • На молочных телятах identity-correct 99,23% против 99,80% у SAM 3.
  • На Edinburgh pigs HOTA 74,05 против 75,54 у SAM 3; на PigTrack — 73,25 против 85,87.
  • Один student на датасет: выбирайте ближайший сценарий или дистиллируйте свой.
  • Нужны собственные веса SAM 3 от Meta и принятие условий доступа facebook/sam3.

Пример

path = hf_hub_download("Sonam5/FarmSAM", "calf/calf_student_night_mined/best_backbone_student_v2.pt")
student, meta = core.load_student(path, device="cuda")
x = core.make_student_transform()(Image.open("frame.jpg").convert("RGB"))
features = student(x[None].to("cuda", torch.bfloat16))
# [1, 1024, 72, 72]
ПодробнееСкорость: 4,83–5,01 fps end-to-end против 3,25 fps у SAM 3 на одной A10, 4 объекта, с декодированием кадров.
Backbone student: 41,56–42,61 мс на кадр против 152,46 мс у SAM 3.
load_student сам выбирает EMA-веса; .pt — pickle, проверяйте SHA-256.
Ограничения: нужны стартовые боксы/маски/детектор; Edinburgh — non-commercial; веса под SAM License; не для трекинга людей.
Discuss Telegram 0

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.

    # Sonam5/FarmSAM
    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 Sonam5/FarmSAM --local-dir ~/models/FarmSAM
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to FarmSAM
  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 FarmSAM (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

Sonam5/FarmSAM
Author
Sonam5
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments