Skip to main content

Voyager SDK

Post your questions, share your expertise, and get support to power your AI projects with the Voyager SDK

  • 29 Topics
  • 134 Replies
29 Topics
lorenzo.lamberti
Cadet
lorenzo.lambertiCadet
asked in Voyager SDK
Firefly ITX-3588J : i can not load the metis board not detected

Hello everyone, I am trying to use the Metis PCIe (4GB) in combination with the firefly ITX-3588J.Issue: I can not load the Metis kernel module (driver) into the Linux kernelSetup: I have the Metis PCIe with 4GB of ram  the Firefly ITX-3588J with 16GB of ram with a RockChip. OS: Ubuntu 22.04.4 LTS. I have a Gen3 Pcie with 4 lanes, details below```        LnkCap:    Port #0, Speed 8GT/s, Width x4, ASPM L0s L1, Exit Latency L0s <4us, L1 <16us            ClockPM- Surprise- LLActRep- BwNot- ASPMOptComp+``` So far:I installed the ubuntu OS as described here: https://support.axelera.ai/hc/en-us/articles/25556437653138-System-Imaging-Guide-Firefly-RK3588 I installed the latest 1.3.3 voyager sdk release.  i can see the metis board from lspci``` firefly@firefly:~$ lspci | grep accelerators01:00.0 Processing accelerators: Axelera AI Metis AIPU (rev 02)```however, i do not detect any drivers for the board, even after installing the sdk. Both the following commands do not return anything:```

152
Spanner
Axelera Team
H
Cadet
HermioneCadet
asked in Voyager SDK
inference.py quick start - import Gst failed ModuleNotFoundError

Hi,My goal is to try out a few tutorials with the latest SDK release. Looks like a python package is not found, any clues what I could try?System: Axelere M.2 Card with Aetina Eval SystemVoyager SDK v1.3.1 - a fresh install of the SDK this timeFirst installed the latest driver https://software.axelera.ai/artifactory/axelera-apt-source/metis-dkms/metis-dkms_1.0.2_all.debthen I ran the installer with./install.sh --all --media --user <email addr> --token <token>I got a few suspicious warnings and hit ‘y’ a number of times. Then install of 186 packages completed.[185/186] Install gstreamer1.0-rtspbuilding operatorsrefreshing pcie and firmware0000:01:00.0 : DeviceDevice 0: metis-0:1:0 1GiB m2 flver=1.2.0-rc2 bcver=1.0 clock=800MHz(0-3:800MHz) mvm=0-3:100%Installation complete, but with unresolved issues (see above) firefly@aetina:~/Documents/g2-testing/voyager-sdk$ source venv/bin/activate I proceeded to try inference.py with one of the included videos as input(venv) firefly@aet

685
M
Ensign
S
Cadet
SiegfriedCadet
asked in Voyager SDK
uv build

Hi,I was wondering if there exist an astral uv (pip alternative) build for the voyager-sdk?This would speed up setup of the sdk and allow easier merging with our development environment.Thanks in advance!

161
Spanner
Axelera Team
V
Cadet
vpc_aiCadet
asked in Voyager SDK
Trying to build the axstreamer and gstaxstreamer components

Hi Team,I’m working with the Voyager SDK (using docker container) and trying to build the axstreamer and gstaxstreamer components. The source files examples reference:#include <axruntime/axruntime.hpp> (examples/axruntime/axruntime_example.cpp)But I can’t find axruntime.hpp anywhere in the SDK.I’m planning to run inference using a webcam as the input on my mac, with a pipeline that includes ffmpeg and mediamtx for streaming.Is axruntime supposed to be part of the SDK repo, a separate dependency, or is there a script I missed?Thanks!Val.

242
V
Cadet
G
Cadet
GiodstCadet
asked in Voyager SDK
Error in deploying WASB model

Hi,I get an error when I want to deploy the WASB model for sports ball detection and tracking, converted to onnx with opset 15:ERROR   : Traceback (most recent call last):ERROR   :   File "<frozen compiler.top_level>", line 584, in quantizeERROR   :   File "<frozen qtools_tvm_interface.graph_exporter_v2.graph_exporter>", line 254, in exportERROR   :   File "<frozen qtools_tvm_interface.graph_exporter_v2.graph_exporter>", line 209, in _convert_operatorsERROR   :   File "<frozen qtools_tvm_interface.graph_exporter_v2.replacement_functions>", line 242, in closureERROR   : RuntimeError: Multiple consecutive arithmetic operations by runtime ops found in the graph. This is not allowed presently.ERROR   : ERROR   : The above exception was the direct cause of the following exception:ERROR   : ERROR   : Traceback (most recent call last):ERROR   :   File "/home/aetina/Desktop/voyager-sdk/axelera/app/compile.py", line 456, in compileERROR   :     the_manifest = top_level.c

442
G
Cadet
G
Cadet
GiodstCadet
asked in Voyager SDK
Yolo11 with custom weights conversion and deploy

Hi,When trying to deploy yolo11 with custom weights, I noticed a significant performance drop due to the model conversion to onnx. Can I ask what method and parameters were used to convert the axelera yolo11s-coco-onnx model?Thank you!

20019
W
Ensign
W
Ensign
WGPravinEnsign
asked in Voyager SDK
How to bind specific video inputs to each model when running parallel + cascading pipelines (inference.py)

Hi ​Axelera  team,I’m running parallel and cascading pipelines together with inference.py—the pipelines initialize and generally work. The issue is input routing: during inference I’m unable to reliably select/bind the correct video (file/RTSP/USB) to each network. For example, launching two independent detectors plus one cascade with multiple media sources (e.g., ./inference.py modelA modelB media/a.mp4 media/b.mp4 --pipe=gst --verbose) often results in the streams attaching to the wrong model or failing to open, while one stream runs as expected.Is there a supported way to explicitly map inputs to models/pipelines when using inference.py? For instance: CLI flags to pin inputs by index/name (e.g., --input[0]=..., --input[1]=...) YAML fields that define per-network input URIs A recommended PipeManager pattern for deterministic source‑to‑network routing when combining parallel and cascading graphs   the command i wrote:./inference.py parellel5_yv11pose_scpmyv8_bcfyv8_yv11_ppeyv8 med

773
W
Ensign
shashi.chilappagari
Ensign
shashi.chilappagariEnsign
posted in Voyager SDK
Axelera Now Supported in DeGirum Cloud Compiler

We’re excited to announce that Axelera is now fully supported in the DeGirum Cloud Compiler, our browser-based tool that lets you compile YOLO models in just a few clicks. No toolchains, no local setup, no guesswork. Just upload your PyTorch checkpoint and receive an Axelera-optimized binary, often in under 10 minutes.Note: A free DeGirum AI Hub account is required to access the compiler and test your models directly in the browser.✅ What It Supports YOLOv8 and YOLO11 All model sizes: n, s, m, l, x Tasks: Object Detection Classification Semantic Segmentation Oriented Bounding Box Detection Keypoint Estimation Custom input resolutions during compilation Integrated C++ postprocessing via DeGirum PySDK Instant in-browser testing — no download required Fast compilation — typically under 10 minutes 🎯 Who This Is ForThis tool is built for developers and researchers who want to: Skip manual toolchain setup Quickly compile PyTorch YOLO models to run on Ax

463
shashi.chilappagari
Ensign
M
Cadet
Manikanta R45Cadet
asked in Voyager SDK
Can I Run Custom Python Scripts on Axelera Metis AI Device Beyond Model Inference?

Hi Axelera Team, I'm exploring the Axelera Metis AI device for edge deployment and had a quick question: Is it possible to run our own custom Python scripts or shell commands on the device Or is the device strictly limited to running AI model inference pipelines through your SDK, without the ability to execute external scripts or commands? Looking forward to your guidance. Thanks in advance!

323
Spanner
Axelera Team
mcunha
Ensign
mcunhaEnsign
asked in Voyager SDK
Support to zsh

I am using zsh as my shell application. While activating the environment, I encountered this error:   voyager-sdk git:(release/v1.3) ✗ source venv/bin/activate _AX_set_env:4: bad substitutionDo you think providing support for shells other than bash would help? 

454
Spanner
Axelera Team
F
Cadet
Falcon9Cadet
asked in Voyager SDK
Metis M.2

Esistono e quali sono le architetture di riferimento ?Oppure dei benchmark ( in caso di tuning 🚗)...A scanso di equivoci e per un Proof-Of-Concept 😎

91
Spanner
Axelera Team
A
Cadet
admyainfotech1Cadet
posted in Voyager SDK
Why Voice Search SEO is the Future of Mobile Marketing

Why Voice Search SEO is the Future of Mobile Marketing1. The Mobile-First Era is Already HereThe mobile revolution has reshaped how we live, work, and interact with technology. Today, most users turn to their smartphones not just to browse, but to speak to search engines. Whether it’s asking Siri for the nearest coffee shop or commanding Google Assistant to play a song, voice interaction has become second nature.With over 70% of internet traffic now coming from mobile devices, optimizing for mobile-first behavior is no longer optional. Consumers are increasingly looking for faster, hands-free ways to access information—making voice search a major player in modern marketing. 2. Why Voice Search is Taking OverVoice search isn’t just a novelty—it’s a necessity. People now expect their phones to understand and respond to them conversationally. This shift is driven by:Convenience: Talking is faster than typing.   Accuracy: Voice recognition accuracy has improved drastically.   Contextual se

50
A
Cadet
G
Cadet
GillesCadet
asked in Voyager SDK
Compiling yolov5 doesn't give correct input/output shapes

I’ve used the yolov5s.pt file provided from the official yolov5 repository and converted it to onnx with this command:  python3 export.py --weights yolov5s.pt --imgsz 640 --batch-size 1 --include onnx --opset 17  . I’ve then tried to compile this with this command:  compile -i /home/axelera/Vision.AxeleraTesting/data/yolov5s.onnx -o /home/axelera/Vision.AxeleraTesting/data/compile --overwriteI got this output:09:49:45 [INFO] Dump used CLI arguments to: /home/vintecc/axelera/Vision.AxeleraTesting/data/compile/cli_args.json09:49:45 [INFO] Dump used compiler configuration to: /home/vintecc/axelera/Vision.AxeleraTesting/data/compile/conf.json09:49:45 [INFO] Input model has static input shape(s): ((1, 3, 640, 640),). Use it for quantization.09:49:45 [INFO] Data layout of the input model: NCHW09:49:45 [INFO] Using dataset of size 100 for calibration.09:49:45 [INFO] In case of compilation failures, turn on 'save_error_artifact' and share the archive with Axelera AI.09:49:45 [INFO] Quantizing

435
J
Axelera Team
S
Ensign
saraEnsign
asked in Voyager SDK
inference on a frame obtained from the memory space

Hi again ​@Spanner ! 😊I need to infer an image obtained from byte data stored in a memory space (https://docs.python.org/3/library/mmap.html) and after converting the bytes I get an image.I've seen the sources available for create_inference_stream and I know that I can make inferences to images (https://github.com/axelera-ai-hub/voyager-sdk/blob/release/v1.2.5/docs/tutorials/application.md), but I have to specify the/path/to/file. But in this case I won't have that information.In the code I'll have this line where I read the bytes in the memory space:        frame_size=image_width*image_height*3 //RGB        frame_bytes = mm.read(frame_size)And then I get the new image:      frame = np.frombuffer(frame_bytes,dtype=np.uint8).reshape((image_height,image_width, 3))Is there any way to infer “frame”?Thank you!

21312
Spanner
Axelera Team
G
Cadet
GiodstCadet
asked in Voyager SDK
Yolo11l custom weigts deploy

Hi, I’m getting very pool performance on Yolo11l custom weigts after the conversion in onnx.I’ve used opset 17 and 16, do you have any suggestions?Thanks.

211
G
Cadet
F
Cadet
fatima-zahra.chenaniCadet
asked in Voyager SDK
axelera-runtime.tar.gz

Bonjour,Je n’arrive pas à récupérer le fichier `axelera-runtime.tar.gz`. Le script `install.sh` affiche “Runtime Installed OK” mais ne télécharge ni n’extrait quoi que ce soit.Aucune URL n’apparaît dans les logs et `libaxruntime.so` est absent de mon système.Merci de me fournir un lien de téléchargement direct pour installer le runtime manuellement.Cordialement,   

442
F
Cadet
lorenzo.lamberti
Cadet
lorenzo.lambertiCadet
asked in Voyager SDK
Voyager-sdk on singularity

Dear all, I am trying to set up the metis PCIe accelerator on a intel workstation. Later we will move to a full risc-v embedded setup.In out lab workstation we are forced to use singularity instead of docker, so i am trying to translate the docker container into a singularity one.I think i am very close to make everything work. However, i am now stuck with a problem i can’t figure how to solve. in particular, i get a weird error when i try to run a simple ./inference.py with one of the pre-trained models. Below the full description on how i create, run the container, and how i run the example from the voyager sdk my `voyager-sdk-1.2.5.def` fileBootstrap: dockerFrom: ubuntu:22.04%labels Author You Version voyager-sdk-1.2.5%environment export DISPLAY=$DISPLAY%post apt-get update && apt-get install -y \ sudo \ git \ pciutils \ lsb-release \ x11-utils curl \ wget \ gnupg2 \ lsb-release \ software-prop

735
P
Cadet
M
Cadet
maximiliankirCadet
asked in Voyager SDK
Yolo Performance Issues

Hello, I try to use a YOLO model on 8 camera streams in parallel on the Metis PCIe card. 20-30 fps per camera stream is all I need. With the 548 fps end-to-end stated for YOLOv8s in https://axelera.ai/metis-aipu-benchmarks, it should be possible to reach ~68 fps per camera. As a small test I wrote a python script, which creates a inference stream with 8 videos as input:from axelera.app import config, display, inf_tracersfrom axelera.app.stream import create_inference_streamdef run(window, stream): for frame_result in stream: window.show(frame_result.image, frame_result.meta, frame_result.stream_id) fps = stream.get_all_metrics()['end_to_end_fps'] print(fps.value)def main(): tracers = inf_tracers.create_tracers('core_temp', 'end_to_end_fps', 'cpu_usage') stream = create_inference_stream( network="yolov5s-v7-coco", sources=[ str(config.env.framework / "media/traffic1_1080p.mp4"), str(config.env.framework / "media/traffic1_

1548
M
Cadet
Z
Cadet
zefirCadet
asked in Voyager SDK
Is graphical video output expected to work from within the docker SDK?

Hi all,I’m a docker noob, so please bear with me if this is a stupid question.To get the SDK operational on my Ubuntu-24.04 system I followed the Install Voyager SDK in a Docker Container and can make it up to test-running some inferences. Alas, while I expected the videos to be displayed in a dedicated window at full resolution, for me they are shown in the active text-console (ncurses-based I assume), obviously coarse and slow.I tried to learn how to activate graphics output from within docker, but all I got was that docker essentially is meant to be text-only and graphics usually is provided via web-browser of pages served from within docker.So to put it bluntly: those videos showing Voyager SDK performing object detection at exhibitions at full-resolution and full-framerate, are those from within docker or are they run from a native ubuntu-22 system?If it is possible to tune docker to have that video output, a mini-howto would help a lot. 

405
Z
Cadet
D
Cadet
doctoreCadet
asked in Voyager SDK
Import custom models

Hi there,I've been truly impressed by this chip over the past couple of days—its object detection and LLM capabilities, combined with its energy efficiency, are remarkable. However, for my current project, I need to use Gemma3 and Whisper. From what I understand, your SDK requires using precompiled paths, which suggests that if a model isn’t officially supported by Axelera AI, we might not be able to run it directly.Am I understanding this correctly, or is there an alternative way to use custom models like Gemma and Whisper within your ecosystem?Thanks in advance!

783
Spanner
Axelera Team
S
Ensign
saraEnsign
asked in Voyager SDK
Raspberry Pi 5 - Metis inference with GStreamer Pipeline

Hi, I'm working on an RPI5 and I have the voyager SDK inside a container with ubuntu 22 and it works. I wanted to try running an inference on a stream using GStreamer. First, I tried to access the stream and I managed with the following command:GST_PLUGIN_PATH=/voyager-sdk/operators/lib AXELERA_DEVICE_DIR=../opt/axelera/device-1.2.5-1/omega LD_LIBRARY_PATH=/voyager-sdk/operators/lib:/opt/axelera/runtime-1.2.5-1/lib:$LD_LIBRARY_PATH gst-launch-1.0 rtspsrc location='rtsp://…..' latency=200 ! rtph264depay ! avdec_h264 ! videoconvert ! autovideosink Now I wanted to try the axinferencenet plugin but I'm having problems. I've tested various commands and formats, such as: GST_PLUGIN_PATH=/voyager-sdk/operators/lib AXELERA_DEVICE_DIR=../opt/axelera/device-1.2.5-1/omega LD_LIBRARY_PATH=/voyager-sdk/operators/lib:/opt/axelera/runtime-1.2.5-1/lib:$LD_LIBRARY_PATH gst-launch-1.0 rtspsrc location='rtsp://...' latency=200 ! rtph264depay ! avdec_h264 ! videoconvert ! video/x-raw,format=BGRA !  axtran

1526
H
Axelera Team
H
Cadet
Hiroyuki KojimaCadet
asked in Voyager SDK
Error Installing Voyager SDK 1.2.5 in Docker

Hi, I followed the "Install Voyager SDK in a Docker Container" guide to try installing SDK version 1.2.5, but I encountered an error with one of the packages during the process.Could you please let me know how to resolve this issue?The attached file contains the installation log. ---    [150/193] Installing development libraries    [150/193] Install axelera_compiler    [150/193] python3 -m pip --disable-pip-version-check install --index-url XXXX@software.axelera.ai/artifactory/axelera-dev-pypi/ --no-deps -c cfg//requirements-ubuntu-2204-amd64.txt "axelera_compiler"    [150/193] Looking in indexes: XXXX:****@software.axelera.ai/artifactory/axelera-dev-pypi/    [150/193] WARNING: 401 Error, Credentials not correct for https://software.axelera.ai/artifactory/axelera-dev-pypi/axelera-compiler/(X)    [150/193] ERROR: Cannot install axelera_compiler because these package versions have conflicting dependencies.    [150/193] The conflict is caused by:    [150/193]     The user requested axeler

1056
Spanner
Axelera Team
F
Ensign
floritEnsign
asked in Voyager SDK
voyager token

it ask me every time about token … yes I use my registered emailand I did not found any token in my accountwtf

251
Spanner
Axelera Team
quant.geek
Cadet
quant.geekCadet
asked in Voyager SDK
Ubuntu 24.04 support, and Docker containers

Seems Ubuntu 24.04 is not supported at the moment. What are the plans for this version? Can you give some insight on what needs to be changed to get this installed on that version?

35421
quant.geek
Cadet
C
Ensign
ChinmayMulgundEnsign
asked in Voyager SDK
Issues Deploying and Running yolov5m6.onnx on M.2

I am encountering issues when trying to deploy and run the YOLOv5m6 ONNX model on the Metis M.2. Below are the details of the steps I followed and the issues that arose:1. Deployment Process:I used the following command to deploy the model:./deploy.py customers/mymodels/yolov5m6-object-detection.yamlThe deployment was successful, but I encountered warnings about memory overflow during the buffer fitting process:LowerTIR failed to fit buffers into memory after iteration 0/4. Pool usage: {L1: alloc:4,195,840B avail:4,194,304B over:1,536B util:100.04%, L2: alloc:6,864,384B avail:8,077,312B over:0B util:84.98%, DDR: alloc:54,275,328B avail:260,046,848B over:0B util:20.87%}2. Inference Process:I then attempted to run inference with the following command:./inference.py c-yolov5m6-object-detection chinmay/Vehicle-CV-ADAS/assets/VID-20250208-WA0008.mp4 --save-output ~/Desktop/voyager-sdk/chinmay/saved_videos/output.mp4The model failed to run due to a ShapeInferenceError related to incompatibl

651
Spanner
Axelera Team

Badge Winners

Show all badges
Code of Conduct

Address

HTC5, High Tech Campus
5656 AE Eindhoven
The Netherlands
Email: info@axelera.ai

Menu

AI Accelerators

AI Software

Metis Evaluation Systems

Partners

Product Inquiry

Company

Our Story

Our Team

Careers

News

Contact

Reducing CO₂ with
Axelera’s Forest

Follow Us

LinkedIn

Twitter

Sign Up for Our Newsletter

EU flag Axelera AI is co-funded by the European Union’s Horizon research and innovation programme.

Privacy Policy  |  Cookie Policy  |  Terms of Sale  |  Vulnerability Disclosure Policy

Copyright 2025 © All rights Reserved.