Multilabel ImageNet-1K: Automated Multi-Label Annotations with Spatial Masks

Automated multi-label annotations for the ImageNet-1K training split include spatial masks for selected object-level labels. The dataset is designed for easy inspection and reuse but does not include the original images. The annotations were created by author k3999 and last updated on June 1, 2026.

Use Cases

• Multi-label image classification training based on automated annotations.
• Object localization analysis based on spatial masks.
• Benchmarking model performance on a multi-label version of ImageNet-1K.
• Inspecting and validating automated annotation methods.

Strengths

• Annotations include spatial masks for object-level labels.
• Annotations are designed for easy inspection and reuse.
• Last updated on June 1, 2026.

Limitations

• Row count, column names, file formats, and license are unknown.
• Original ImageNet-1K training images are not included and must be sourced separately.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

ImageNet-1K training split.

Collection Method

Automated annotation.

Freshness

Last updated 2026-06-01 23:55:05.