HumanSignal / HumanSignal/label-studio-ml-backend
Trouble using segment_anything_2_image in Label Studio
- Dominant language
- Python
- Stars
- 1.1k
- Forks
- 490
- Avg merge
- 1d 14h
- Merged PRs (30d)
- 3
Description
Hi HumanSignal team
First off, I really appreciate the open-source version of Label Studio. Amazing work!
Unfortunately, I haven’t been able to get SAM2 for image segmentation working, even after following the steps from your guide here:
🔗 [segment_anything_2_image example](https://github.com/HumanSignal/label-studio-ml-backend/tree/master/label_studio_ml/examples/segment_anything_2_image)
My implementation is available here:
👉 https://github.com/mxochicale/labelstudio
I thought the problem might be related to:
* LABEL_STUDIO_URL and LABEL_STUDIO_API_KEY, but fixing those didn’t help
* I also tried adding showInline="true" to the labeling configuration
* The ML backend logs look fine (though I’m not sure if I’m missing something):
```
ml-backend | [2025-10-25 00:14:03,387] [DEBUG] [hydra.core.utils::set::279] Setting JobRuntime:name=UNKNOWN_NAME
ml-backend | [2025-10-25 00:14:03,387] [DEBUG] [hydra.core.utils::set::279] Setting JobRuntime:name=app
ml-backend | [2025-10-25 00:14:05,293] [INFO] [root::_load_checkpoint::174] Loaded checkpoint sucessfully
ml-backend | [2025-10-25 00:14:10,488] [DEBUG] [label_studio_ml.api::log_request_info::185] Request headers: Host: localhost:9090
ml-backend | User-Agent: curl/7.81.0
ml-backend | Accept: */*
ml-backend |
ml-backend |
ml-backend | [2025-10-25 00:14:10,488] [DEBUG] [label_studio_ml.api::log_request_info::186] Request body: b''
ml-backend | [2025-10-25 00:14:10,489] [DEBUG] [label_studio_ml.api::log_response_info::191] Response status: 200 OK
ml-backend | [2025-10-25 00:14:10,489] [DEBUG] [label_studio_ml.api::log_response_info::192] Response headers: Content-Type: application/json
ml-backend | Content-Length: 41
ml-backend |
ml-backend |
ml-backend | [2025-10-25 00:14:10,489] [DEBUG] [label_studio_ml.api::log_response_info::193] Response body: b'{"model_class":"NewModel","status":"UP"}\n'
```
Here’s what I see in the UI:
Do you have any recommendations or know what I might be missing?
Thanks a lot for your help!
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with the linked segment_anything_2_image example and compare it with the reporter's implementation at github.com/mxochicale/labelstudio. Review the Label Studio configuration, LABEL_STUDIO_URL, LABEL_STUDIO_API_KEY, and the backend request logs to identify where image segmentation stops working. Done means documenting a reproducible cause and a confirmed working configuration, but the issue currently gives no specific code location or completion test.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-vision, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100