AI-Based Representation of Figure-Ground Segregation from Occlusion Stimuli

Supplementary file 1_Approaching human visual perception through AI-based representation of figure-ground segregation.docx is a 25.3 KB document by Chakkai Yip, last updated March 18, 2026. It contains results from training multiple convolutional neural networks on simple overlapping occlusion stimuli to probe border-ownership inference. The analysis examines performance under degraded contour conditions and the role of junction-like configurations.

Use Cases

• Benchmarking feedforward neural network architectures on border-ownership inference tasks based on occlusion stimuli.
• Analyzing the contribution of geometric context versus isolated edges in visual segregation based on junction configurations.
• Studying the hierarchical progression from local edge responses to spatially coherent features in AI models of perception.

Strengths

• Dataset is openly licensed under CC-BY-4.0.
• Research methodology is described, involving training multiple CNN architectures on occlusion stimuli.
• File was updated on 2026-03-18, indicating recent activity.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Dataset is very small at 25.3 KB, indicating limited scope, likely containing summary results or figures rather than raw experimental data.

Provenance

Source

Chakkari Yip via figshare.

Collection Method

Results from training and testing convolutional neural networks on occlusion stimuli.

Time Range

null

Freshness

Last updated 2026-03-18 07:43:19; freshness should be verified.

Geography

null