EyePCR Benchmark For Ophthalmic Surgical Video Analysis

EyePCR is a benchmark dataset for fine-grained perception, knowledge comprehension, and clinical reasoning in ophthalmic surgery. It is derived from the OphNet large-scale video benchmark and was created by author EvergreenTree, with the dataset last updated in May 2025.

Use Cases

• Classify surgical phases using video frames from ophthalmic procedures.
• Predict instrument presence and usage from video segments.
• Generate descriptive captions for surgical actions and steps based on video content.
• Train models for anomaly detection during specific surgical workflow stages.

Strengths

• Derived from the OphNet benchmark, a large-scale collection of ophthalmic surgery videos.
• Designed for multiple complex tasks including fine-grained perception and clinical reasoning.

Limitations

• Specific row count, video duration, and number of procedures are unknown.
• Geographic origin and demographic diversity of the patient data are unspecified.

Provenance

Source

Derived from the OphNet video benchmark.

Collection Method

null

Time Range

null

Freshness

Dataset page was updated on 2025-05-16.

Geography

null