Published February 11, 2026 | Version v1
Annotation collection Open

EchoVisuALL - benchmark dataset

  • 1. ROR icon Helmholtz Zentrum München

Description

Here, we provide the expert-curated benchmarking data for EchoVisuALL. The ZIP folder includes a CSV file 'expert_benchmarks.csv' and the annotated frames as cropped PNGs with a size of 1024x344 pixels.

The CSV file contains the following  columns: 

Column Description
file_id File identifier (1-20)
file_name Name of the corresponding file
frame_number Number of the frame within the DICOM
x_coords X-coordinates of the annotation mask (comma-separated list, -1 if bg=1)
y_coords Y-coordinates of the annotation mask (comma-separated list, -1 if bg=1)
an_type The annotations are stored as a polygon with x and y coordinates. 
bg This column indicates whether there are no annotations within this frame (bg=1).
annotator Number of annotatos (1-5). The aggregates mask is labeled by 'agg'.

Files

expert_annotations.zip

Files (213.7 MB)

Name Size Download all
md5:495f93c5734a1ce349afa947b64fb24f
213.7 MB Preview Download