Skip to content

Instantly share code, notes, and snippets.

@simonw

simonw/bbox.md Secret

Created August 15, 2026 20:37
Show Gist options
  • Select an option

  • Save simonw/a05cc78b2061555bd61d3bb9686e689f to your computer and use it in GitHub Desktop.

Select an option

Save simonw/a05cc78b2061555bd61d3bb9686e689f to your computer and use it in GitHub Desktop.

2026-08-15T20:28:01 conversation: 01m03hraf82yak6cjqd15825d9 id: 01m03hrafbby12wxr38teb4jc4

Model: lmstudio/qwen/qwen3.8-27b (resolved: qwen/qwen3.8-27b)

Prompt

Return JSON bounding boxes for the pelicans in this photo, 0-1000 scale for each dimension

Attachments

  1. image/jpeg: https://static.inaturalist.org/photos/714731804/large.jpg

Reasoning

The user wants bounding boxes for the pelicans only. I need to identify the pelicans in the image, distinguishing them from the smaller seabirds (cormorants/gulls).

Looking at the image, there are two large pelicans with long bills and gular pouches standing on rocks. The other birds are smaller and darker.

Pelican 1 (left): standing on a rock, long bill open. Roughly x from 195 to 370, y from 290 to 780.

Pelican 2 (center-right): long bill closed, body angled. Roughly x from 445 to 675, y from 320 to 850.

Let me refine coordinates in 0-1000 scale. The image dimensions appear to be standard landscape. I'll estimate based on relative positions.

Left pelican: head top around y=290, feet around y=780. Left edge of body ~195, right edge of bill ~370.

Right pelican: head top ~320, feet ~850. Left edge of body ~445, right edge of bill ~675.

I'll provide these two boxes.

Response

[
	{"bbox_2d": [195, 290, 370, 780], "label": "pelicans"},
	{"bbox_2d": [445, 320, 675, 850], "label": "pelicans"}
]

Token usage

749 input, 353 output, {"completion_tokens_details": {"reasoning_tokens": 272}}

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment