The Computer Vision API

Easily experiment with the best algorithms in Computer Vision and Deep Learning thanks to Ikomia's Open Source Python API


Time saving

Computer Vision API

You can now use the best Computer Vision algorithms on your own images/videos in a few lines of code :

  • No waste of time looking for the best detection, classification, pose or image recognition algorithms: we select and test them for you.
  • A plug & play API to easily execute any algorithm.
  • No need to worry about upgrades anymore : Ikomia manages your algorithms’ Python dependencies.

An API for all developer profiles

Junior developer

You're new to Computer Vision and Deep Learning?

Easy learning

Ikomia API is the ideal tool to learn Computer Vision and Deep Learning through practice. Test the best algorithms in a few lines of code and create your first workflows effortlessly.

Senior developer

You have good knowledge of Computer Vision and Deep Learning?

A portable low-code tool

Create and integrate your own Python algorithms, then mix them with other tested algorithms. Deploy your algorithms on any computing server (Google Colab, AWS, GCP...) in a few lines of code.

Easy to use

Simple installation

You can install Ikomia API through the PIP package management system.

Our API is compatible with Python 3.7 / 3.8 / 3.9 .

					pip install ikomia

Free access to Ikomia HUB

Join the Ikomia community and access our hub of state-of-the-art 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"



With just a few lines of code.

					from ikomia.utils import ik
from ikomia.dataprocess import workflow

# Create your worflow
wf = workflow.create("My workflow") 

# Add algorithms to your workflow
yolo_id, yolo = wf.add_task(ik.infer_yolo_v5) 
canny_id, canny = wf.add_task(ik.ocv_canny) 

# Connect your algorithms
wf.connect_tasks(wf.getRootID(), yolo_id) 
wf.connect_tasks(yolo_id, canny_id) 

# Run

# YOLO output image with bounding boxes
img_detect = wf.get_image_with_graphics(yolo_id)
# Canny output image
img_final = wf.get_image(canny_id, index=0)

cv2.imshow("Detection", img_detect)
cv2.imshow("Result", img_final)

# Export your workflow for reuse"/path/to/my_workflow_name.json")

800+ models

in our Ikomia HUB, selected and tested 

Open source

for easy sharing with the community

CV specialized

with more than 15 years of experience

800+ algorithms

in our library of models, selected and tested

Open source

for easy sharing with the community

CV specialized

with more than 15 years experience


Use cases



Simple workflow

How to make a simple workflow

Neural Style transfer

How to run Neural Style transfer on your images


How to train and run YOLO v7 on your datasets

Detectron2 Object Detection

How to use Detectron2 Object Detection


How to use MMPose

Coming soon


Fichier 2

Python API documentation

Learn and create your first Ikomia Python app with our API. Enjoy the Ikomia tools.


Ikomia GitHub

Ikomia API and all our models are open source. Visit our GitHub repo for more information.