If you decided to start experimenting with Ikomia Computer Vision API, you are at the right place. For those of you who want to go solo, well we built it clean and simple so just help yourself with our documentation and our Github, and enjoy the experience!
In the team, we use the API on a daily basis to deploy our workflows, but also to launch trainings in a few lines of code on any machine.
If you are searching for a step by step guide to come on board, this content is for you.
First steps with Ikomia API
What is it? A Python Open Source Computer Vision API. It’s plug & play, and icing on the cake it manages your algorithms’ Python dependencies.
What does it do? It executes any algorithm, it creates workflows, and it deploys your algorithms on any computing server (Google Colab, AWS, GCP…).
Where do I get the algorithms? This is where the magic begins : join the Ikomia Community to choose the algorithms that match your needs from our Open Source Ikomia HUB. More that 70 algorithms and 800 models have been selected and tested for you, and the team adds more each week.
You will find the top-notch algorithms from your preferred sources : OpenCV, Openmmlab… But you can also create and integrate your own Python algorithms and mix them with other tested algorithms.
How do I start ? First you need to install the API:
pip install ikomia
Then access the HUB if you want to pick one or several algorithms:
import ikomia import os # Easy and unsafe authentication | Only for personal use os.environ['IKOMIA_USER'] = "your_username" os.environ['IKOMIA_PWD'] = "your_password" ikomia.authenticate()
Then you can follow this step by step tutorial to easily prototype your first simple workflow. You will use the YOLOv5 algorithm in order to detect objects in an image and then apply the stylization filter on the detected objects.
It’s a good start!
We aim at covering the whole spectrum of Computer Vision, whether it is classification, object detection, object segmentation, pose estimation or generative algorithms. Just browse the HUB to see what fits your goal.
Eager to test something else ? We recommend that you take a look at this other tutorial: How to use Detectron 2 object detection.
A step further : training and integration
Now that you know how to prototype a Computer Vision workflow with the API, it’s time to dive into Deep Learning. We will use the object detection star YOLO V7 to easily create your own object detector thanks to Ikomia API. Just follow the Colab and apply it to your own use case.
We’re always happy to take a sneak peek at your productions. You can share them by mail at firstname.lastname@example.org and tell us whether you would like a feedback or if you wish to share with the community (or both!).
And remember: your questions get answered on the Ikomia community Q&A discussion.
Be innovative with Ikomia!