Test Images and Training Videos for Motion-Induced Color Illusion Models

4.4 GB of multimedia resources support a 2025 BioRxiv study on subjective color perception. Eiji Watanabe and colleagues published test images, training videos, code, and initial model weights for their artificial neural network models. These models generate artificial subjective color to investigate the underlying mechanisms of color perception illusions.

Use Cases

• Training predictive neural network models based on the provided training videos
• Testing model performance on color illusions using the published test images
• Reproducing computational frameworks for subjective color perception research
• Investigating the mechanisms of motion-induced color illusions using the initial model weights

Strengths

• Includes 4.4 GB of multimedia and code resources directly supporting a published BioRxiv preprint
• Released under a permissive CC-BY-4.0 license for open reuse and modification
• Contains both training data (videos) and evaluation data (images) for model development

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

Eiji Watanabe

Collection Method

Created for research on predictive networks and color illusions.

Freshness

Last updated 2026-05-11 06:20:25