skills.rest
xjtulyc
xjtulycCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

image-analysis-art

Extract CNN embeddings and color/texture descriptors from painting images.

33|6|Updated Mar 17, 2026
Search Tags:#color palette#grad-cam#art analysis#cnn features#wikiart#style classification#texture analysis
One-click install
npx skills add https://github.com/xjtulyc/awesome-rosetta-skills --skill image-analysis-art
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: image-analysis-art
Source: https://github.com/xjtulyc/awesome-rosetta-skills/tree/main/skills/16-art-music/image-analysis-art
Command: npx skills add https://github.com/xjtulyc/awesome-rosetta-skills --skill image-analysis-art

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires torch, torchvision, opencv-python, numpy, matplotlib, scikit-learn, umap-learn, datasets, scikit-image, pillow, and includes references (resource) and assets (resource) components.

What problem does it solve?

It helps you analyze paintings computationally by turning raw artwork images into measurable style, color, and texture signals you can compare or cluster.

Core Features & Use Cases

  • CNN feature extraction & visualization: Extract ResNet50/VGG16 embedding vectors and use PCA/UMAP or t-SNE to explore style clusters.
  • WikiArt-driven style classification: Fine-tune a pretrained CNN to predict WikiArt style/artist/genre labels and create attribution via Grad-CAM.
  • Color + texture quantification: Compute dominant palettes (k-means), compare color distributions (HSV histograms with Bhattacharyya distance), and characterize brushstroke texture using Gabor filterbanks and rotation-invariant LBP.

Quick Start

Use the image-analysis-art Skill to extract ResNet50 features from a folder of WikiArt paintings and cluster them to identify style groupings.

Frequently Asked Questions about image-analysis-art

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

FAQPage Schema
How do I extract CNN features from painting images for style classification?▼

You can compute a dominant color palette from paintings using k-means clustering, and compare color distributions across artworks by calculating HSV histograms with Bhattacharyya distance.

What's the best way to quantify brushstroke texture in artwork computationally?▼

The best way to quantify brushstroke texture in artwork is by characterizing the images using Gabor filterbanks alongside rotation-invariant Local Binary Patterns (LBP) descriptors.

Can I use Grad-CAM for attribution in WikiArt style prediction?▼

Yes, you can use Grad-CAM for attribution by fine-tuning a pretrained CNN to predict WikiArt style, artist, and genre labels from your painting images.

Do I need PyTorch and torchvision to cluster art images with PCA and UMAP?▼

Yes, you need PyTorch and torchvision for deterministic CNN feature extraction, and you can then apply PCA or UMAP dimensionality reduction to visualize and cluster the art style embeddings.

How does CNN embedding similarity search work for computational art history?▼

CNN embedding similarity search works by extracting ResNet50 or VGG16 vector representations from painting images, allowing you to computationally compare and cluster measurable style signals.

Related Skills

Explore More Education & Research Skills→
zhangzhang-111-izhangzhang-111-i
2

senior-computer-vision

Guide object detection and image segmentation with PyTorch and ONNX deployment.

Community
Advanced
erichowenserichowens
181

clip-aware-embeddings

Match images to text descriptions using CLIP-compatible embeddings for semantic search.

Community
Intermediate
SPIRAL-EDWINSPIRAL-EDWIN

pca-analyzer

Perform PCA dimensionality reduction and feature extraction on numeric DataFrames.

Community
Intermediate
nahisahonahisaho
3

scientific-image-analysis

Automates microscopy image segmentation, morphometric measurement, and texture analysis via Python libraries.

Community
Advanced