CSAILVision / CSAILVision/ADE20K

Typographical and labeling inconsistencies found in ADE20K object names (e.g., “school desk”, “sckool desk”, “scool desk”, etc.)

Open
#53 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Jupyter Notebook
Stars
433
Forks
66
PR merge metrics
No merged PRs in 30d

Description

Dear ADE20K team,

First, thank you for maintaining and openly sharing such a valuable dataset for the vision community. I am currently using ADE20K for research purposes, and while exploring the annotations, I noticed a few typographical inconsistencies in some object names that may affect label consistency and downstream analyses.

For example, the following entries appear to correspond to the same concept (“school desk”) but with different spellings:

| name | name_ndx |
|------|-----------|
| sckool desk | 3196 |
| scool desk | 3197 |
| sscool desk | 3198 |
| schoo desk | 3199 |

These seem to be minor typographical variants that could be unified under the correct label `"school desk"` to improve consistency in object-level statistics and model training.

Could you please confirm whether there is a preferred channel or contribution process to help identify and correct these inconsistencies (e.g., through a pull request, structured report, or CSV correction proposal)?

I would be glad to assist in compiling a list of similar issues across the dataset if that would be helpful.

Thank you for your time and for your outstanding work on ADE20K.

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.