huggingface-vision-trainer

Fine-tune computer vision models on Hugging Face Jobs infrastructure.

2|Updated Mar 29, 2026

Search Tags:#fine-tuning #vision #training #segmentation #huggingface #object-detection

Ready to run

npx skills add https://github.com/NagyVikt/codex-plugins --skill huggingface-vision-trainer-nagyvikt

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: huggingface-vision-trainer
Source: https://github.com/NagyVikt/codex-plugins/tree/main/plugins/hugging-face/skills/vision-trainer
Command: npx skills add https://github.com/NagyVikt/codex-plugins --skill huggingface-vision-trainer-nagyvikt

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires transformers, datasets, monai, trackio, torch, torchvision, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill removes the complexity and hardware barriers associated with training custom computer vision models, allowing you to fine-tune state-of-the-art architectures on cloud GPUs without local infrastructure.

Core Features & Use Cases

Multi-Task Support: Fine-tune models for object detection (D-FINE, RT-DETR), image classification (timm models), and segmentation (SAM/SAM2).
Production-Ready Workflows: Includes automated dataset validation, cost estimation, and Trackio monitoring to ensure training jobs succeed and results are saved to the Hub.
Use Case: Quickly fine-tune a lightweight RT-DETR model on a custom dataset of industrial parts to enable real-time defect detection on edge devices.

Quick Start

Use the huggingface-vision-trainer skill to validate my dataset username/my-dataset and then launch a training job for an object detection model on a t4-small GPU.

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 on Hugging Face Jobs?▼

To fine-tune a vision model on Hugging Face Jobs, you can use this skill to validate your dataset, estimate hardware costs, and launch automated training for object detection, classification, or segmentation tasks. It handles model persistence to the Hub automatically.

Can I train a segmentation model using SAM2 with Hugging Face Transformers?▼

Yes, you can train a segmentation model using SAM2. This skill integrates Hugging Face Transformers and the monai library to execute training and evaluation for segmentation networks directly on cloud GPUs.

Does this training workflow support custom datasets for object detection?▼

Yes, this training workflow supports custom datasets for object detection. It includes automated dataset validation to ensure your data is correctly formatted before launching the fine-tuning job on architectures like D-FINE or RT-DETR.

What is the best way to estimate GPU costs for fine-tuning computer vision models?▼

The best way to estimate GPU costs for fine-tuning computer vision models is using this skill's hardware-specific cost estimation feature. It calculates expected expenses before launching your training job on Hugging Face Jobs infrastructure.

How do I monitor training jobs and save results to the Hugging Face Hub?▼

You can monitor training jobs and save results to the Hugging Face Hub by utilizing the integrated Trackio monitoring and automated model persistence features. This ensures your fine-tuned checkpoints are safely stored after training completes.