huggingface-vision-trainer

Automate vision-model training and fine-tuning on Hugging Face Jobs.

Updated Aug 23, 2026

Search Tags:#object-detection #image-classification #huggingface-jobs #hub-push #vision-training #sam-segmentation

Ready to run

npx skills add https://github.com/domattioli/DomI --skill huggingface-vision-trainer-domattioli

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: huggingface-vision-trainer
Source: https://github.com/domattioli/DomI/tree/main/plugins/huggingface-skills/skills/huggingface-vision-trainer
Command: npx skills add https://github.com/domattioli/DomI --skill huggingface-vision-trainer-domattioli

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires transformers>=5.2.0, accelerate>=1.1.0, albumentations>=1.4.16, timm, datasets>=4.0, torchmetrics, pycocotools, trackio, huggingface_hub, evaluate, scikit-learn, torchvision, monai, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill enables end-to-end training and fine-tuning of vision models on Hugging Face Jobs cloud GPUs, covering object detection, image classification, and SAM/SAM2 segmentation workflows.

Core Features & Use Cases

End-to-end vision training: fine-tune DETR/RT-DETR, timm-based classifiers, and SAM2 segmentation with cloud GPUs.
Dataset prep & evaluation: COCO-format preparation, Albumentations augmentations, and mAP/mAR evaluation; Hub persistence and Trackio monitoring.
Cost estimation & hardware guidance: provides hardware flavors, cost estimates, and timeout recommendations; enables hub push to share results.

Quick Start

Train a small vision-detection workflow on Hugging Face Jobs by running a ready-made training script that fine-tunes a chosen model and pushes the result to the Hub.

Frequently Asked Questions about huggingface-vision-trainer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I fine-tune a vision model for object detection on Hugging Face Jobs?▼

Fine-tuning a vision model for object detection on Hugging Face Jobs involves running automated training scripts that configure DETR or RT-DETR with COCO-format data, apply Albumentations augmentations, and evaluate mAP/mAR before pushing weights to the Hub.

Can I train SAM2 segmentation models using Hugging Face cloud GPUs?▼

Training SAM2 segmentation models using Hugging Face cloud GPUs is fully supported through automated workflows that handle dataset preparation, environment configuration with required dependencies like transformers and accelerate, and secure Hub persistence.

What is the best way to prepare COCO-format datasets for image classification training?▼

Preparing COCO-format datasets for image classification training is handled natively by the workflow, which uses datasets and albumentations libraries to structure inputs, apply augmentations, and ensure compatibility with timm-based classifiers.

Does Hugging Face Jobs training support automatic model pushing to the Hub?▼

Hugging Face Jobs training supports automatic model pushing to the Hub using secure HF_TOKEN authentication, allowing trained vision models to be shared and persisted directly from the cloud GPU environment after evaluation.

Do I need to configure hardware flavors and cost estimates manually for vision training on Hugging Face Jobs?▼

Configuring hardware flavors and cost estimates manually for vision training on Hugging Face Jobs is guided by the workflow, which provides specific hardware recommendations, cost projections, and timeout settings to optimize cloud GPU resource allocation.

Why does my vision training workflow require accelerate and transformers dependencies?▼

Vision training workflows require accelerate and transformers dependencies because they provide the foundational framework for distributed training, model optimization, and handling architectures like DETR and SAM2 on Hugging Face Jobs cloud infrastructure.