.svg?alt=media&token=65d6f32d-0e72-4040-80b2-8e858896c71d)
Get started
Deploy your first model in just 5min using Syndicai Platform.
Yolo V5 model prediction
You just need 5min to go through the quick start. No need to prepare your own model, no need to have access to any cloud, and no need to fork the repository.
In this very quick tutorial, you will deploy one of the most popular object detection algorithms YoloV5 with just a click of a button.
Impossible? Let's do it!
- 1.Go to the YoloV5 repository, and click the Syndicai Deploy button placed in the model's description. You will be redirected to the Syndicai Platform.
- 2.Press Add to confirm the form and add a model to the Syndicai platform. The platform will connect to the repository and add your YoloV5 model to the platform.
- 1.In the next step, press the Deploy button which is in the top, right corner of Model Profile.
- 2.The platform will open Add a new deployment form since there are no deployments.
- 3.Fill in name of the deployment e.g. Production, Fill in the branch of the deployment e.g. master, and press Add in order to create a new deployment. You will be redirected to the Deployment profile
- 4.You can see that status of the deployment is Starting. In order to explore what is happening under the hood, please change the tab from Overview to Releases, and press the recent release by clicking on #1. You will see logs of your starting deployment.
- 5.After a couple of minutes, your deployment will have the status Running.
- 1.With the Running status of your deployment go to the Overview tab in the Deployment Profile, and scroll down to Validate & Integrate sections.
- 2.Press the icon of the pencil in order to input some sample data (the box with the text: Add an example input for your model in JSON format.)
- 3.Place the sample input JSON e.g.
{ "url": "https://images.pexels.com/photos/2083866/pexels-photo-2083866.jpeg?auto=compress&cs=tinysrgb&h=750&w=1260" }, and click Update. - 4.Click Send Request and everything will be ok if your response is as follows
200 /9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHB...
- Before any deployment, we need to have a trained model. For the purpose of the following tutorial, we took a deep learning model that was already trained and prepared for deployment so there was no need to add/modify anything in the code.
- In the first step of the whole deployment process, we need to connect a git repository with a model by providing the URL to the repository and path to syndicai.py. However, when we clicked the Syndicai Deploy button a platform automatically recognized a git repository and filled the form for us providing the URL, and path of that model.
- The next step was about the actual deployment of the model. By clicking the Deploy button the platform asks us to first create a Deployment by filling the form with the name, and the branch. Form submission not only created a deployment but also fires the new release. Each release consists of two steps:
- Building - a model is wrapped with a webservice and packaged with the docker container.
- Starting - a ready container with your model was served on the cloud.
If you found that tutorial helpful or would like to learn more, then send us a message on slack or by mail!
Take any model that is already trained, add a syndicai.py with requiremnts.txt and deploy it on the platform. You can learn more about the whole preparation phase in the Develop section.
Create new deployments e.g. staging, dev, and click Upgrade to run a new release of your deployments.
Go to our repository with prepared models ready to deploy in order to play a little more and explore how they are prepared so they can be easily deployed.
Go to Deployment Profile and scroll to Validate & Integrate section in order to access code snippets of your deployment.
With generated API you can create a Showcase website based on your deployed models. In order to do it quickly, you can use our boilerplate and easily deliver your AI model in a stylish way.
Last modified 1yr ago