Personix-Octo: 8,355 Multi-Theme Images for Classification

Personix-Octo is a multi-theme image classification dataset containing 8,355 JPEG images across 9 distinct visual categories. Created by Poralus and updated in March 2026, the collection provides approximately 850 images per theme for benchmarking computer vision models. The dataset supports high-resolution workflows, including 4K resolution options.

Use Cases

• Training image classifiers to distinguish between 'anime' and 'nature' themes
• Benchmarking model performance on 'abstract' visual patterns
• Testing high-resolution 4K image processing pipelines

Strengths

• 8,355 total images
• 9 distinct visual themes
• Balanced distribution with ~850 images per category
• Supports 4K resolution workflows

Limitations

• Small sample size for deep learning at 8,355 total records
• Limited to 9 specific thematic categories
• Unknown image provenance may include label noise or synthetic content

Provenance

Source

Poralus

Freshness

Last updated March 2026.

Geography

United States