Skip to main content
Answer

Problem with Voyager SDK 1.5.1

  • December 1, 2025
  • 7 replies
  • 98 views

Forum|alt.badge.img

I’ve already redeployed the model:
./inference.py yolo11m-custom dataset --show-device-fps --no-display --pipe gst
arm_release_ver: g13p0-01eac0, rk_so_ver: 9
INFO    : Using dataset val
WARNING : Using deprecated heuristic transpose detection. Redeploy the model to include explicit output info
Segmentation fault

Best answer by Habib

Hi ​@Giodst 

Can you please try to use this [1] yaml for your custom model? and redeploy, i.e. please make sure old artifacts are completely removed from the build/ dir. Thanks!

---

[1] https://github.com/axelera-ai-hub/voyager-sdk/blob/release/v1.5/ax_models/zoo/yolo/object_detection/yolo11m-coco-onnx.yaml

7 replies

Spanner
Axelera Team
Forum|alt.badge.img+2
  • Axelera Team
  • December 1, 2025

Hi ​@Giodst! Just to double check - when you say you’ve redeployed the model, did you also recompile it? There’s a breaking change in Voyager 1.5 that requires models to be recompiled.

I’m guessing you have, as you know what you’re doing, but I thought it best to check first as we dig deeper 👍


Forum|alt.badge.img
  • Axelera Team
  • December 1, 2025

Hi ​@Giodst,

Could you also try using the precompiled yolo11m model from axdownloadmodel? If the model-zoo version works with inference.py, it would help confirm that the SDK installation is not the source of the issue.

Many thanks!


Forum|alt.badge.img
  • Author
  • Ensign
  • December 2, 2025

Hi ​@Habib the precompiled model works.


Spanner
Axelera Team
Forum|alt.badge.img+2
  • Axelera Team
  • December 2, 2025

Thanks for letting us know ​@Giodst ! That’s really helpful, especially on a new release. Glad it’s working again!


Forum|alt.badge.img
  • Author
  • Ensign
  • December 3, 2025

But for custom recompiled model the error remains: WARNING : Using deprecated heuristic transpose detection. Redeploy the model to include explicit output info
Segmentation fault


Forum|alt.badge.img
  • Axelera Team
  • Answer
  • December 3, 2025

Hi ​@Giodst 

Can you please try to use this [1] yaml for your custom model? and redeploy, i.e. please make sure old artifacts are completely removed from the build/ dir. Thanks!

---

[1] https://github.com/axelera-ai-hub/voyager-sdk/blob/release/v1.5/ax_models/zoo/yolo/object_detection/yolo11m-coco-onnx.yaml


Forum|alt.badge.img
  • Author
  • Ensign
  • December 3, 2025

It works, the only difference is the addition of this entry in the pipeline: 

inference:

handle_all: false

Many Thanks! ​@Habib ​@Spanner