Skip to content
Other

fashionpedia-yolo26n-3classes

a lightweight model for segmenting clothing in photos

2 min readupdated

This checkpoint is fine-tuned for instance segmentation (detecting each object individually and tracing its exact contour) of clothing items in photos. For every garment detected, the model outputs its class, a confidence score, a bounding box, and a pixel-level mask. Built on the compact Ultralytics YOLO26n-seg architecture, it was fine-tuned by Qudef1 on 10% of the Fashionpedia training split as part of the FashionSwap AI lab project. It is well-suited for developers building virtual try-on apps or local image editing tools that need to cleanly isolate specific garments on a person.

Key features

  • Predicts the class, bounding box, and precise pixel-level mask for each clothing item in an image.
  • Built on the lightweight Ultralytics YOLO26n-seg architecture and fast enough to run even on a CPU.
  • Trained on images at an input resolution of imgsz=512.
  • Stores the class list directly inside the checkpoint, accessible via model.names.

What you can do

  • Automatically generate garment masks for local inpainting, swapping, or recoloring clothes with image editing models.
  • Plug the model into a wardrobe cataloging or e-commerce pipeline to detect and crop individual clothing items from photos.
  • Use it as a fast baseline for benchmarking against other clothing detectors and segmentation models.

Example

from huggingface_hub import hf_hub_download
from ultralytics import YOLO

weights = hf_hub_download(repo_id="Qudef1/fashionpedia-yolo26n-3classes", filename="best.pt")
model = YOLO(weights)
result = model.predict(source="person.jpg", device="cpu", imgsz=512, conf=0.25, retina_masks=True)[0]
result.save(filename="prediction.jpg")
Read moreDetails
Trained on 10% of the Fashionpedia training split at 512 px resolution; the repository includes the best.pt and last.pt checkpoints, the args.yaml config, and per-epoch metric logs in results.csv.
Limitations: Since this is a lab baseline, validate the masks on your own data before deploying to production—inaccuracies can occur along fine edges or when garments overlap.
Runs through the standard ultralytics library via either Python or the command line.
Examples
Discuss Telegram 1 84

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

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.

    # Qudef1/fashionpedia-yolo26n-3classes
    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 Qudef1/fashionpedia-yolo26n-3classes --local-dir ~/models/fashionpedia-yolo26n-3classes
    # 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 fashionpedia-yolo26n-3classes (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

Qudef1/fashionpedia-yolo26n-3classes
Author
Qudef1
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

841

Ask AI about this model

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

Comments