Egomask: Ego-Vision Mask Annotation Benchmark

Egomask is a benchmark dataset for mask annotation in egocentric vision. The dataset's structure includes separate directories for training data and benchmark evaluation subsets. It was uploaded by XuuuXYZ and last updated on August 4, 2025.

Use Cases

• Benchmarking segmentation model performance based on the provided benchmark structure
• Training models for object tracking in egocentric video based on the 'egotracks' directory
• Developing models for referential expression grounding based on the 'refego' directory
• Evaluating model performance on sequences of varying length based on the 'short', 'medium', and 'long' subsets

Strengths

• Provides a structured benchmark for evaluation, as indicated by the detailed folder hierarchy
• Includes separate training and benchmark datasets, suggesting a clear split for model development and testing

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

huggingface

Freshness

Last updated 2025-08-04 15:24:38