Quick Start Guide: Seeed Studio reCamera with Ultralytics YOLO26
reCamera was introduced for the AI community at YOLO Vision 2024 (YV24), Ultralytics' annual hybrid event. It is mainly designed for edge AI applications, offering on-device AI processing and straightforward deployment.
With support for diverse hardware configurations and open-source resources, it is well suited to prototyping and deploying computer vision solutions at the edge.

Why Choose reCamera?#
The reCamera series is purpose-built for edge AI applications, tailored to the needs of developers. Its key features include:
-
RISC-V Powered Performance: At its core is the SG200X processor, built on the RISC-V architecture, delivering efficient performance for edge AI tasks. With the ability to execute 1 trillion operations per second (1 TOPS), it handles tasks like real-time object detection.
-
Optimized Video Technologies: Supports advanced video compression standards, including H.264 and H.265, to reduce storage and bandwidth requirements without sacrificing quality. Features like HDR imaging, 3D noise reduction, and lens correction maintain image quality, even in challenging environments.
-
Energy-Efficient Dual Processing: While the SG200X handles complex AI tasks, a smaller 8-bit microcontroller manages simpler operations to conserve power, making the reCamera ideal for battery-operated or low-power setups.
-
Modular and Upgradable Design: The reCamera is built with a modular structure, consisting of three main components: the core board, sensor board, and baseboard. This design allows developers to swap or upgrade components as projects evolve.
Quick Hardware Setup of reCamera#
Follow the reCamera Getting Started page and select your device (such as reCamera 2002 Series, Gimbal, or HQ POE) to open its Quick Start Guide for initial onboarding of the device such as connecting the device to a network and accessing the Node-RED web UI for quick previewing of detection results.
Inference Using Pre-installed YOLO Models#
reCamera comes pre-installed with four Ultralytics YOLO models, and you can choose your desired model within the Node-RED dashboard.
Step 1: If you have connected reCamera to a network, enter the IP address of reCamera on a web browser to open the Node-RED dashboard. If you have connected the reCamera to a PC via USB, you can enter 192.168.42.1. The detection model is loaded by default.

Step 2: Click the green circle at the bottom right corner to access the Node-RED flow editor.
Step 3: Click the model node and click On Device.

Step 4: Choose one of the four pre-installed models and click Done. This example selects the pose model.
Step 5: Click Deploy and when it finishes deploying, click Dashboard.

The dashboard now shows the pose estimation model running on the camera feed.

Export to cvimodel: Converting Your YOLO26 Model#
If you want to use a custom-trained YOLO26 model with reCamera, follow the steps below.
The conversion first exports a PyTorch model to ONNX, then converts it to the MLIR model format, and finally converts MLIR to cvimodel to run inference on-device.
Export to ONNX#
Export an Ultralytics YOLO26 model to ONNX model format.
Installation#
To install the required packages, run:
pip install ultralyticsFor detailed instructions and best practices related to the installation process, check our Ultralytics Installation guide. While installing the required packages for YOLO26, if you encounter any difficulties, consult our Common Issues guide for solutions and tips.
Usage#
from ultralytics import YOLO
# Load a YOLO26 model
model = YOLO("yolo26n.pt")
# Export the model to ONNX format
model.export(format="onnx", opset=14) # creates 'yolo26n.onnx'For more details about the export process, visit the Ultralytics documentation page on exporting.
Export ONNX to MLIR and cvimodel#
After obtaining an ONNX model, refer to the Convert and Quantize AI Models page to convert the ONNX model to MLIR and then to cvimodel.
Real-World Applications of reCamera#
reCamera's on-device computer vision and modular design make it suitable for a wide range of real-world scenarios.
-
Fall Detection: Designed for safety and healthcare applications, the reCamera can detect falls in real time, making it well suited to elderly care, hospitals, and industrial settings where rapid response is critical.
-
Personal Protective Equipment Detection: The reCamera can support workplace safety by detecting PPE compliance in real time. It helps identify whether workers are wearing helmets, gloves, or other safety gear, reducing risks in industrial environments.

-
Fire Detection: The reCamera's real-time processing capabilities make it well suited to fire detection in industrial and residential areas, providing early warnings to prevent potential disasters.
-
Waste Detection: It can also be used for waste detection in environmental monitoring and waste management with YOLO11.
-
Car Parts Detection: In manufacturing and automotive industries, it aids in detecting and analyzing car parts for quality control, assembly line monitoring, and inventory management.

FAQ#
To set up your reCamera for the first time, follow these steps:
- Connect the reCamera to a power source
- Connect it to your network using the reCamera Getting Started page (select your device to open its Quick Start Guide)
- Access the Node-RED web UI by entering the device's IP address in a web browser (or use
192.168.42.1if connected via USB) - Start using the pre-installed YOLO models immediately through the dashboard interface
Yes, you can use custom-trained YOLO26 models with reCamera. The process involves:
- Export your PyTorch model to ONNX format using
model.export(format="onnx", opset=14) - Convert the ONNX model to MLIR format
- Convert the MLIR to cvimodel format for on-device inference
- Load the converted model onto your reCamera
For detailed instructions, refer to the Convert and Quantize AI Models guide.
- Export your PyTorch model to ONNX format using
Unlike traditional IP cameras that require external hardware for processing, reCamera:
- Integrates AI processing directly on the device with its RISC-V SG200X processor
- Offers 1 TOPS of computing power for real-time edge AI applications
- Features a modular design allowing for component upgrades and customization
- Supports advanced video technologies like H.264/H.265 compression, HDR imaging, and 3D noise reduction
- Comes pre-installed with Ultralytics YOLO models for immediate use
These features make reCamera a standalone solution for edge AI applications without requiring additional external processing hardware.
Comments50
Ask a question, share your results, or help someone else.
tcpipchip
we wrote a book and trainned to yolo26 on ultralytics...i recall that works
Glenn JocherUltralyticsYes, a custom-trained
YOLO26model can run onreCamerausing theONNX -> MLIR -> cvimodelworkflow in the reCamera guide; if you hit a blocker, please share the exact step and full error here so the Ultralytics team can help.
HanJammer
This site mentions recamera comes with YOLO26 (which is why I bought two of them) yet YOLO11 is installed (which is visible on screenshots as well) and I can't load YOLO26 - how to do that? I can't find reliable guide :(
HanJammer
Ah nevermind, I think I got it, still it would be better not to advertise the camera "reCamera comes pre-installed with four Ultralytics YOLO26 models" which is basically not true.
HanJammer
I whipped out some 'single click' tool to convert YOLO26 onnx-es which makes it a bit easier... I need to test it a bit, for now it supports the base model. Perhaps I will push it to my github later this week.
tcpipchip
This is fantastic.
When you finish it, i will add to the BOOK.
HanJammer
Here is the first version of the converter util (it's just GUI and few scripts over the actual workflow). For now it was tested with a base detection model under Windows. More to come.
https://github.com/HanJammer/reCamera_YOLO26
Video presentation here: https://youtu.be/XgPjyj2U8YQ
tcpipchip
Here is the first version of the converter util (it's just GUI and few scripts over the actual workflow). For now it was tested with a base detection model under Windows. More to come.
https://github.com/HanJammer/reCamera_YOLO26
Video presentation here: https://youtu.be/XgPjyj2U8YQ
wow!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
i will test soon!
Glenn JocherUltralyticsThanks for sharing this; for anyone testing, the supported path today is still the reCamera integration guide together with Seeed’s model conversion guide, and we’ll review the docs wording around preinstalled models since some devices appear to ship with YOLO11 rather than YOLO26.
tcpipchip
@glenn-jocher Have you a recamera ?
Glenn JocherUltralyticsI don’t personally keep a reCamera on my desk, but the workflow in the Seeed Studio reCamera guide is the supported path; if you have a specific setup or export question, please drop it here and the Ultralytics team can take a look.
tcpipchip
I tested on recamera with my trainned pose keypoint model... Same problem...no keypoint
Need to test on yolo8 on tdl sdk
Glenn JocherUltralyticsThanks for checking this; for custom pose models on reCamera, please drop the exact export and conversion steps here in the discussion so the Ultralytics team can take a look, since keypoints should work only after a correct
ONNX -> MLIR -> cvimodelpipeline as outlined in the reCamera guide and Seeed’s conversion guide, and it’s worth first confirming your exportedONNXmodel still produces keypoints before converting.
tcpipchip(edited)
did you do calibration ? where the images to me to the calibration :(
Glenn JocherUltralyticsCalibration isn’t done in the Ultralytics export step; for reCamera, you export to ONNX first with the reCamera guide, then calibration happens during Seeed’s ONNX→MLIR→
cvimodelflow using a small representative set of your own images, as described in Seeed’s Convert and Quantize AI Models guide. If you hit an error in that step, please drop the full log here and we’ll take a look.
tcpipchip
still waiting...
tcpipchip
send me your cvimodel too...yolo11 and yolo8
tcpipchip
please, send me arguments used in your scripts...
tcpipchip(edited)
did you train keypoints on animals and persons, right ?
tiruvijayakumar
NOO, i didnt train on animals, i trained on these classes for pose detection for persons using yolov11n-pose.pt weights. I fine‑tuned the YOLOv11n‑pose.pt model on my custom classes:
Running Walking Golf Skate Horse riding Kicking Diving I just uploaded animal detection model to justify that custom model deployment with quantization for object detection(yolo11n.pt) on recamera has no issues, it was another cvi model which i had no issues, it worked PERFECTLY. i only issue what i am facing is the pose detection quantized model where i am not able to see the keypoints on the inference output but bounding box is drawn along with predicted class label refer below:
now just trying to train the model for pose detection using yolov8 and then i would try quantizing the model and check if the keypoints are shown in the inference output
Glenn JocherUltralyticsThanks, that helps narrow it down — since detection
cvimodels work and only the quantized pose model loses keypoints, this looks more like a pose conversion or postprocessing issue in the ONNX → MLIR →cvimodelstep than a training problem. Please keep the details here so the Ultralytics team can see them, and share the exact conversion commands and full text logs from the reCamera guide and Seeed model conversion guide, plus whether the same pre-quantized ONNX shows keypoints before conversion.tiruvijayakumar(edited)
i ran predictions on ultralytics hub on the .pt model format, i attached the whole problem scenario and image content :https://github.com/orgs/ultralytics/discussions/22090?sort=new#discussioncomment-16346206 Any ways uploading the prediction on ultralytics hub in this thread also :
so on the model.pt file inference shows this data, this confirms that pose estimation model works. { "images": [ { "shape": [ 426, 640 ], "speed": { "preprocess": 1.41517, "inference": 130.14456, "postprocess": 0.99221 }, "results": [ { "name": "Kicking", "class": 5, "confidence": 0.80929, "box": { "x1": 0.26068, "y1": 0.10738, "x2": 0.55882, "y2": 0.85426 }, "keypoints": { "x": [ 0.41923001408576965, 0.42656999826431274, 0.4166699945926666, 0.45181000232696533, 0.42857998609542847, 0.488209992647171, 0.43380001187324524, 0.543470025062561, 0.41604000329971313, 0.5516700148582458, 0.4164600074291229, 0.46198999881744385, 0.4427199959754944, 0.357340008020401, 0.427839994430542, 0.29197001457214355, 0.43891000747680664 ], "y": [ 0.2315099984407425, 0.21907000243663788, 0.21740999817848206, 0.21602000296115875, 0.21322999894618988, 0.2783600091934204, 0.2824600040912628, 0.3255299925804138, 0.35097000002861023, 0.32879000902175903, 0.3619599938392639, 0.5179499983787537, 0.5104399919509888, 0.6517800092697144, 0.6389099955558777, 0.7586100101470947, 0.7324900031089783 ], "visible": [ 0.9994300007820129, 0.9990299940109253, 0.9990000128746033, 0.999019980430603, 0.9988300204277039, 0.9995700120925903, 0.9996600151062012, 0.998960018157959, 0.9992700219154358, 0.9985499978065491, 0.997439980506897, 0.9995700120925903, 0.9995700120925903, 0.9988800287246704, 0.9989299774169922, 0.982990026473999, 0.9852499961853027 ] } } ] } ], "metadata": { "imageCount": 1, "functionTimeAlive": 78321.917, "functionTimeCall": 0.138, "model": "best.pt", "version": { "ultralytics": "8.4.30", "torch": "2.11.0+cpu", "torchvision": "0.26.0+cpu", "python": "3.14.3" } }, "networkTime": 1.214 }
I didnt capture the conversion commands and text logs during the onnx->mlir->cvimodel, would repeat the same steps and paste the images in this thread itself.
tiruvijayakumar
Exported the yolov11n-pose.pt(finetuned) on google colab to onnx format
Uploaded the model on netron to view the structure, onnx is in the format IRv9 ,opset v20(would downgrade it). Classes are running, walking, golf-swing, skateboarding, riding-horse and kicking
Seeed studio guide for downgrading
Initial structure of workspace folder
Step1: pull sopho/tpuc_dev:v3.1
Step2: starting the container
Step3: installing onnx
Step4: Clone Downgrade Script
Step5: Downgrade Script settings IR v8 and Opset v17
Step6: Install TPU-MLIR tools inside container:
Step7: clone tpu-mlir
Step8: setup environment tpu-mlir(long step)
Step9: according to seeed studio guide, i have to include outputs before the model's head as the parameter for this particular step model_transform
--model_name yolo11n
--model_def yolo11n.onnx
--input_shapes "[[1,3,640,640]]"
--mean "0.0,0.0,0.0"
--scale "0.0039216,0.0039216,0.0039216"
--keep_aspect_ratio
--pixel_format rgb
--output_names "/model.23/cv2.0/cv2.0.2/Conv_output_0,/model.23/cv3.0/cv3.0.2/Conv_output_0,/model.23/cv2.1/cv2.1.2/Conv_output_0,/model.23/cv3.1/cv3.1.2/Conv_output_0,/model.23/cv2.2/cv2.2.2/Conv_output_0,/model.23/cv3.2/cv3.2.2/Conv_output_0"
--test_input ../image/dog.jpg
--test_result yolo11n_top_outputs.npz
--mlir yolo11n.mlir i have the onnx file, could someone of the team help me identify the outputs before the model's head? @glenn-jocher here is the onnx file https://drive.google.com/file/d/1UF25DK2zXx0GCWdcx55SPYHL71SylDeg/view?usp=sharing
tcpipchip
and some animal pictures...