bytedance / bytedance/UI-TARS

Wrong coordinates for very big screen images

Open
#150 2 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
11.5k
Forks
877
PR merge metrics
No merged PRs in 30d

Description

Hello,
We observed UITARS 1.5 7B giving wrong coordinates for big screen images (3840x2100). When we resized the images to be smaller it worked correctly.

What do you recommend for the max image sizes? Thank you.

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reproducing the coordinate error with the reported 3840x2100 images and compare the results with resized versions. Trace the image-input and coordinate-output path to identify where large dimensions affect behavior; done means the supported maximum image size is established or the issue is reliably reproduced with a focused test.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai, computer-vision
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.