Adding YOLO v5 to our Edge Impulse Support
We are excited to announce a further integration with Edge Impulse: we are adding support for the Edge Impulse YOLO v5 architecture.
Why Scailable + Edge Impulse?
We have been collaborating for some time with the amazing team at Edge Impulse. Edge Impulse makes advanced AI / ML model training effortless. And, due to their focus on edge targets the resulting models (pipelines, really) are computationally extremely efficient. Hence, you can run very capable trained models on relatively small hardware.
With Scailable, we are building the go-to layer for effortless remote deployment (and management) of the full AI/ML pipeline on selected edge devices. Take any edge device with our AI manager pre-installed and you will be able to remotely configure the full AI pipeline and deploy models without having to tailor your solution to the selected hardware. Remote AI deployment saves valuable development time and lets you focus on the solution instead of the hardware platform. And, by providing easy ways of capturing data in context — and feeding this data back to (for example) Edge Impulse for model training — we speed up the AI/ML training cycle allowing you to iterate faster and deliver highly capable solutions.
So, train your models using Edge Impulse, and remotely deploy your solution using Scailable to supported devices. AI/ML solutions build from start to end without any nitty gritty coding on the device.
Why would I want the YOLO v5 integration?
We started out by supporting Edge Impulse’s awesome FOMO model: see our docs here, and see a demonstration of the FOMO model in action on the Advantech ICAM during Embedded World last month here. FOMO brings object detection to highly constrained devices.
However, we sometimes have more resources available, and we want to provide bounding boxes (as opposed to centroids) of the object that we are detecting. Enter YOLO; YOLO is easily trained using Edge Impulse (and using transfer learning, easily retrained for specific purposes). Overall, it’s a proven model architecture that can be made to work for a variety of vision tasks.
Okay, okay, I get it. How does it work?
As stated, creating and deploying a YOLO model using Scailable + Edge Impulse to any Scailable supported edge device is a breeze. We start by creating an Edge Impulse project and training our model:
Next, we connect our Scailable account with the Edge Impulse project to add the trained model to our available model library:
And, finally (What? Already? Yes. 😀) we remotely configure the Scailable AI manager on the selected edge device (which in most cases comes pre-installed) to generate inferences:
You can find a simple, step-by-step, tutorial here.
Where to go from here?
Okay, so creating edge AI solutions from start to end takes minutes. And, iterating based on new training data is a breeze. This is very cool.
However, there is more to come. We currently support model imports from a variety of platforms (Tensorflow, PyTorch, ONNX, etc.) for effortless deployment on any Scailable supported edge device. And, with our recent integration with Edge Impulse, things became even easier. In the next months we will be extending our Edge Impulse integration to cover all Edge Impulse models for effortless deployment using the AI manager (and yes, also extending beyond vision models).
So, stay tuned.
What can I do?
- Well, you can get started using our simple tutorial.
- Rather get a demo? Book a time slot.
- Want to see things in action in the real word? Visit us during the Hannover Messe (17-21 April) in hall 17, stand A52 (with Edge Impulse) or in hall 15, stand D52 (with Advantech).
- Or, immediately want to test on supported hardware? Find a reseller.
- Finally, if you need help building your edge AI solution, or if you want to add your hardware to our supported hardware, contact us.
Scailable supporting Seeed NVIDIA devices
We are excited to announce support for Seeed’s NVIDIA Jetson devices. The AI manager, and all our edge AI development tools, can now readily be used on Seeed devices. As we all know, edge AI solutions often start from creative ideas to optimize business processes with AI. These ideas evolve into Proof-of-Concept (PoC) phases, where […]
From your CPU based local PoC to large scale accelerated deployment on NVIDIA Jetson Orin.
Edge AI solutions often start with a spark of imagination: how can we use AI to improve our business processes? Next, the envisioned solution moves into the Proof-of-Concept (PoC) stage: a first rudimentary AI pipeline is created, which includes capturing sensor data, generating inferences, post-processing, and visualization of the results. Getting a PoC running is […]
Allxon and Scailable Unite to further Revolutionize Edge AI Deployment
In a groundbreaking development that promises to reshape the landscape of edge AI deployment, Allxon and Scailable have joined forces to unveil the Scailable AI Manager, a cutting-edge edge AI deployment middleware now available as a plugin in the Allxon Plugin Station. This strategic collaboration brings together the expertise of Allxon in image-based over-the-air (OTA) […]