How To Annotate and Manage Data With Kili Technology

Kili technology is a data annotation and labelling platform that helps AI companies transform their businesses, with an industrial-level and communicative platform, to manage training datasets to deploy their machine learning applications faster.

Kili technology is a data annotation and labelling platform that helps AI companies transform their businesses, with an industrial-level and communicative platform, to manage training datasets to deploy their machine learning applications faster. Kili Technology was launched in 2018 by François-Xavier Leduc and Edouard d’Archimbaud and headquartered in Paris, France and other two branches in the USA and Singapore. 

Kili Technology offers customizable and intuitive interfaces to address the appropriate machine learning use cases, for all kinds of multimedia text, image, video(DICOM Images, 3D images, and scans) and audio. This is all possible by using ML with active learning, online learning, and semi-supervised learning, and get the labelling process in speed by connecting models to pre annotate the data and monitor accuracy with quality checks. 

They have access for organisations to work in their technical teams or outsource annotation companies to ease the entire data science project and workflow management at both levels, training and production. Provides SaaS(Software as a Service) or On-Premise(Data and Enterprise) workflow with privacy and security. Kili Technology is made accessible to dockerized microservices hosted on Kubernetes. The frontend GUI is made of React, and the backend uses GraphQL and for storage PostgreSQL database. Also has options to drag and drop data from cloud service providers. They have an open-source community in GitHub.

Workflow Management

Kili technology helps in scaling up data science projects during training and production. In every project, there are four major roles – Admin, Labeller, Reviewer, Reader. The admin has access to create, edit, add or delete data, export data, put labels, add members, and control their operations. The annotator will annotate the data provided to him, see the KPIs, and visualize overall work. He cannot modify anything in the project configuration or the imported data, or view other annotators metrics. He has viewed the overview tab, the Review of work and the annotation interface. The reviewer can review and make changes in the labels. He has no access to the Settings tab. The reader has permission to only view the annotated tasks like 3rd proofreading. However, he can navigate over the annotation interface and overview tab.

The consensus is a parameter that measures the compatibility of more than one annotations on the same asset(data), done by several annotators. It makes the system consistent among the annotators with the best data quality for the project. For quality assurance, Kili Technology uses Honeypot that measures the quality against ground truth between the annotation made by an annotator and an asset pre-annotated. 

Image Annotation

The image annotation tools available are points, polyline, polygon, bounding boxes, and segmentation tools that allow performing labelling as per classification(binary or multiclass, to manage ontologies there is hierarchical classification) and object detection and to easily perform OCR(Optical character recognition – image to text). Supported file types are jpg and png.

Object detection

Video Annotation

Features available in the video annotation tool include bounding boxes, segmentation, polyline, point, polygon allowing streamless navigation through frames for object detection(in single and multiple frames), object classification(single and multiple frames), video transcription. Supported files are mkv and mp4.

Medical Imagery(Tumor Detection)

Audio Annotation

Speech to text or Audio transcription to classify audio for identification of speakers or topics over audio and video files. Modify playback speed or timestamps with follow-ups to transcription rhythm. Pre-defined voice transcriptions are present to accelerate annotation. Accepted file formats include mp3, mp4, FLAC.

Text Annotation

Kili technology’s NLP solutions provide rapid text annotation services. Conversational Bot training, entity extraction from text documents be it emails, medical reports, etc. Accepted formats are txt and PDF.

Kili Playground 

Kili Playground is a Python client GraphQL API allowing to control Kili Technology from an IDE.


pip install kili


from kili.authentication import KiliAuth
from kili.playground import Playground
kauth = KiliAuth(api_key='MY API KEY')
playground = Playground(kauth)

CAR Brand Identification

project_example = {
    'title': 'Porsche or Tesla recognition',
    'description': 'Identify and locate cars',
    'input_type': 'IMAGE',
    'json_interface': {
        "jobs": {
            "JOB_0": {
                "mlTask": "OBJECT_DETECTION",
                "tools": [
                "instruction": "What car brand ?",
                "required": 1,
                "isChild": False,
                "content": {
                    "categories": {
                        "TESLA": {"name": "Tesla"},
                        "FERRARI": {"name": "Ferrari"}
                    "input": "radio"
'assets_to_import': [
    'json_response': {
        "JOB_0": {
            "annotations": [{
                "boundingPoly": [{
                    "normalizedVertices": [
                        {"x": 0.16, "y": 0.82},
                        {"x": 0.16, "y": 0.32},
                        {"x": 0.82, "y": 0.32},
                        {"x": 0.82, "y": 0.82}
                "categories": [{"name": "TESLA", "confidence": 100}],
    'model_name': 'car-brand-localisation-v0.0.1'
project = playground.create_empty_project(user_id=os.getenv('KILI_USER_ID'))
    external_id_array=['ex1', 'ex2'])


Carrefour, ArcelorMittal, EDF, VitaDX, Bureau Veritas, Societe Generale, République Française.

More Great AIM Stories

Jayita Bhattacharyya
Machine learning and data science enthusiast. Eager to learn new technology advances. A self-taught techie who loves to do cool stuff using technology for fun and worthwhile.

More Stories


8th April | In-person Conference | Hotel Radisson Blue, Bangalore

Organized by Analytics India Magazine

View Event >>

30th Apr | Virtual conference

Organized by Analytics India Magazine

View Event >>

Yugesh Verma
All you need to know about Graph Embeddings

Embeddings can be the subgroups of a group, similarly, in graph theory embedding of a graph can be considered as a representation of a graph on a surface, where points of that surface are made up of vertices and arcs are made up of edges

Yugesh Verma
A beginner’s guide to Spatio-Temporal graph neural networks

Spatio-temporal graphs are made of static structures and time-varying features, and such information in a graph requires a neural network that can deal with time-varying features of the graph. Neural networks which are developed to deal with time-varying features of the graph can be considered as Spatio-temporal graph neural networks. 

Vijaysinh Lendave
How to Evaluate Recommender Systems with RGRecSys?

A recommender system, sometimes known as a recommendation engine, is a type of information filtering system that attempts to forecast a user’s “rating” or “preference” for an item. In this post, we will look at RGRecSys, a library that performs constraint evaluation of recommender systems.

3 Ways to Join our Community

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Telegram Channel

Discover special offers, top stories, upcoming events, and more.

Subscribe to our newsletter

Get the latest updates from AIM