skills.rest
hung-phan
hung-phanCOMMUNITY
Explore all skills by this author →
GitHub

vision

Recommend vision transformer architectures and fine-tuning steps for classification, detection, and segmentation tasks.

1|Updated Jun 4, 2026
Search Tags:#fine-tuning#classification#detection#model selection#segmentation#vision transformer
One-click install
npx skills add https://github.com/hung-phan/ml-skills --skill vision-hung-phan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: vision
Source: https://github.com/hung-phan/ml-skills/tree/main/skills/ml-review/references/ml-architectures/vision
Command: npx skills add https://github.com/hung-phan/ml-skills --skill vision-hung-phan

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides expert guidance on selecting and fine-tuning the right vision transformer architectures for your classification, detection, or segmentation tasks, saving time and improving accuracy.

Core Features & Use Cases

  • Model Selection: Offers a comprehensive guide to choose the best vision transformer for your specific task, considering speed, versatility, and accuracy.
  • Fine-Tuning: Provides detailed instructions on how to fine-tune ViT-family models with minimal adaptation.
  • Use Case: If you're working on a real-time detection project and need a balance between speed and accuracy, this Skill will guide you to the optimal model and provide code examples for fine-tuning.

Quick Start

Analyze your vision task with the vision skill and get the recommended model and fine-tuning steps for your specific needs.

Frequently Asked Questions about vision

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

FAQPage Schema
How do I choose the best vision transformer for classification, detection, or segmentation tasks?▼

To choose the best vision transformer for classification, detection, or segmentation, evaluate architectures based on speed, versatility, and accuracy. This Skill assesses your task constraints and recommends the optimal ViT-family model for your project.

How do I fine-tune a ViT-family model with minimal adaptation?▼

To fine-tune a ViT-family model with minimal adaptation, follow detailed instructions for adapting pre-trained weights to your dataset. This Skill provides step-by-step fine-tuning guidance and code examples for classification, detection, and segmentation tasks.

What is the best way to balance speed and accuracy for real-time detection using vision transformers?▼

Balancing speed and accuracy for real-time detection using vision transformers requires selecting an architecture optimized for inference latency. This Skill analyzes your real-time constraints and guides you to the optimal model with code examples for fine-tuning.

Do I need prior knowledge of machine learning principles to use vision transformer architectures?▼

Yes, you need prior knowledge of vision transformers and machine learning principles to effectively use this guidance. This Skill provides expert model selection and fine-tuning instructions for ViT-family architectures, assuming foundational ML competency.

Related Skills

Explore More Software Engineering Skills→
ovachieverovachiever
52

clip

Encode images and texts to compute cosine similarity for zero-shot classification and cross-modal retrieval.

Community
Advanced
ovachieverovachiever
52

transformers

Load, infer, and fine-tune pre-trained transformer models across modalities.

Community
Advanced
jorgeasaurusjorgeasaurus
12

finetuning

Fine-tune machine learning models on Azure AI Foundry with Python scripts.

Community
Advanced
hung-phanhung-phan
1

cnn

Design and implement convolutional neural networks for image tasks using PyTorch.

Community
Advanced