Data Collection and Annotation Services for AI and ML Models
Images, videos, speech, text, you name it: we collect and annotate custom training data tailored to your AI needs on an end-to-end basis.
Forget about learning how to use complicated data labeling platforms, communicating with crowd workers or validating data on your own
Get the right ML-data with a white-glove service
Images
Data annotation and labeling
Classification, bounding box annotation
Text
Classification, named entity recognition
Videos
Speech
Object recognition and detection, classification
Audio transcription in 150+ languages, time code annotation
Testing
ASR testing
Usability testing
Images
Speech
Videos
Global data collection
Serge Kuznetsov, CEO
We take care of all the state-of-the-art data collection and annotation under the hood so you can focus on turbo-charging your AI initiatives.
Why Zapisano?
Quick start and rapid scaling
Time is money: The faster you get your first data set, the sooner you can prove your concept and move on to the next iteration.
An international crowd of more than 1 million people is instantly available for most data collection and annotation tasks, and it’s rapidly scalable on the fly.
Details under control
Our project management and tech teams work closely together to make sure no important details slip through the cracks.
In our experience, starting with a clear, thorough project brief and ensuring the process goes as planned is what makes a difference in terms of costs and turnaround times.
Clockwork
Saturn
Deep and versatile expertise
We have completed more than 360 projects, from simple labeling to highly complicated tasks involving offline collection and multistep annotation.
We understand the nuances of each project type and the potential pitfalls to avoid. For the most complex and unique tasks, we build ad hoc teams of data collectors, assessors and analysts.
Swiss knife
Cases
Speech annotation
Transcribed and annotated 2,700 hours of livestreamed video in four languages to train AliExpress’s simultaneous video translation service.
Speech data collection
Collected 12,800 hours of speech data (over 5.7 million utterances) in five languages to develop a voice assistant for VinFast, the Vietnamese carmaker.
Image collection
Collected and labeled 25,000 live images of text from 15 countries to train the Live Text feature in Apple’s iOS.
- Transcription and Emotion Classification of Call Center Employees ConversationsWe transcribed recordings of conversations between call center employees and clients and classified emotions into five categories, tagging specific emotional cues throughout the text. The data was used to develop an employee emotion recognition system.
Client: Real estate rental, purchasing, and valuation service
Volume: 600 hours
Fragment length: 4 minutes, segmented into 10-15 second intervals - Panoramic Video Recording of Apartment InteriorsWe located and filmed the interiors of 20 apartments (studios, one-, and two-bedroom) using a 360-degree camera. The footage was used to develop an MVP system for creating 3D models of residential spaces from a single video.
Client: Innovation department of a digital ecosytem company
Volume: 20 apartments - Transcription of Railway Dispatcher CommunicationsWe transcribed recordings of railway dispatchers' communications to identify irregular or dangerous situations in real-time.
Client: Railway transport company
Volume: 150 hours
Fragment length: 30 seconds, segmented into 5-10 second intervals - Transcription of Booking Managers’ ConversationsWe transcribed conversations of ticket booking department employees to monitor errors and track conflict situations.
Client: Airline
Volume: 100 hours
Fragment length: 5 minutes, segmented into 5-10 second intervals - Transcription of Customer Service DialoguesWe transcribed customer service conversations from an appliance retail chain’s service department, recorded via audio badges, with verbatim text and voice identification.
Client: Appliance retailer service department
Volume: 300 hours
Fragment length: 5 minutes, segmented into 10-15 second intervals - Complex Video Recording of London LandmarksWe filmed the Tower Bridge, Piccadilly Street, and Trafalgar Square in London using specialized equipment according to the client's technical guidelines. The videos were recorded from different spots and used for developing an augmented reality application.
Client: Innovation department of a digital ecosytem company
What would you like to do?
What would you like to do?