Collecting and marking datasets of any complexity
Geodata is a reliable partner in the development and implementation of neural networks. We solve any problems in audio, video, images, text and medicine in a short time
{ Medical Data }
{ Text }
{ Images }
{ Audio }
{ Video }
Measuring work
specific indicators
and openly share the results
with our clients
{ Experience in numbers }
300 000+
Accurate data were collected in the shortest possible time
200 000+
Dialogs were generated of varying complexity
1 m +
Offer images were marked
Geodata company cooperates with 28 leading companies in the CIS and Russia
{ Our partners }
{ Our solutions }
Data markup
Audio
Voice assistants and phonetics: stress, punctuation, transcription. Deciphering children’s speech and incorrect pronunciation.
Video
We will detect moving objects, trace their trajectories, predict movement and classify them according to the necessary characteristics.
Images
We examine the emotional state, select the necessary objects and add tags with comments.
Text
We will place accents and punctuation, and perform a transcription of the text. Let’s create a ready-to-use bot. We recognize the language and translate it and classify the text according to the task.
Own platform
97% markup accuracy
1 million+ marked-up data
{ Marking of radiographs, CT and MRI }
Working with DICOM files in 3D Slicer
Segmenting pathologies
Classification of diseases
Identification of pathologies
5,000+
MRI studies
2,000+
CT studies
1,000+
X-rays
Segmenting pathologies
Classification of diseases
Identification of pathologies
Data collection
Audio
We collect audio data in any amount and format, in different languages, different external conditions, categories of users with transcription and annotation
Video
We work using our experience, knowledge and resources that are necessary for high-quality video data collection. We will prepare data from CCTV cameras in the premises, outdoors, road photography, biometrics, video simulating various scenarios. Will clean and structure the data.
Images
Documents, people, food, cars, unmanned aerial vehicles, etc.
100,000 audio recordings
1000 videos
350,000 photos
Dataset preparation
Audio
Video
Images
Text
Medical images
1000 videos
100,000 audio recordings
350,000 photos
{ Markup cost }
Depending on your goals, you can choose the most suitable tariff
Object detection from about
о.8 r
per object
Marking polygons from
3 r
per object
Semantic markup from
6 r
per object
3D marking of point clouds from
13 r
per object
Skeletal markings
8 r
per object
Audio transcription from
40 r
per object
By time Assessor’s base rate
from 400 RUR/hour
It is possible to organize a work schedule that is convenient for you, including round-the-clock duty of assessors or marking upon request.
For permanent assessors who will work full time, that is, 8 hours on all weekdays, the cost will be
from RUB 64,000/month
If you need specialists with specialized knowledge or need more powerful equipment for work, the price is calculated
If you know exactly the required volume of markings, then we can prepare an individual calculation of the cost of your project.
Price includes specialist fee, workplace, taxes and fees
Each project has a permanent manager who monitors quality.
Replacing a specialist on a project takes 1 day.
{ About the company }
Geodata is a reliable partner in the development and implementation of neural networks. We solve any cases in audio, video, images, text and medicine in a short time
A unique data annotation platform has been developed that provides a flexible approach to collection and annotation. We specialize in confidential data and can effectively engage large numbers of markers. Our 50 experienced employees ensure maximum quality markings, working in the office on a permanent basis. Over the past 5 years, we have optimized business processes, allowing quality control online. Collaboration with 10+ specialized doctors made it possible to collect and tag medical data. Control your data with confidence.
Ruslan S.
Co-founder «Geodata»
350,000 +
Biometric samples collected
5,000 +
5,000 MRI images collected
10,000 +
Person surveyed for medical data
{ Operating principles }
Our staff
Experts in the field of artificial intelligence and computational linguistics, with deep knowledge and professionalism for their work. Our staff improves the quality and speed of our projects
Quality control
We carry out testing and quality control at every stage of cooperation. Our curators closely monitor the process, which ensures efficiency and speed of task completion
Transparency of processes
We provide complete transparency of processes for our clients. We develop pipelines that allow you to monitor the progress of work, ensuring the smooth functioning of our team
Data security guarantee
We ensure complete security of your data. Our employees sign non-disclosure agreements, including NDA, NCA and NSA, and all know-how developed for you remains yours forever
{ Our tools }
{ Work experience }
FinTech
TelecomTecg
Telecom
Machine Learning
Russian financial conglomerate, the largest universal bank in Russia and Eastern Europe Task
Task
Create communication variations for artificial intelligence
Solution
More than 2,000,000 unique chat bot responses have been written
#communication #data collection #chatbot #artificial #intelligence
Russian provider of digital services and services
Task
Collection of data for the quality control library of biometric samples
Solution
More than half a million biometric presentations collected faces and voices. Including presentations with special, intentional performed, distortions in quality.
#biometric#samples #data collection #intelligence #faces
Russian company providing telecommunications, digital and media services in Russia, Armenia and Belarus Task
Task
Collection of audio data for training KWS and ASR models for the device (speaker)
Solution
Collected more than 1000 hours of recordings from different categories of users
#training #recordings #data collection #intelligence #audited #KWS #ASR
Research company in the field of computer vision, data analytics and robotics
Task
Video data for system training "Smart City"
Solution
Video material has been collected containing scenes of simulated use of weapons, fights, forgotten things
#training #recordings #data collection #intelligence #video data
{ Work stage }
Statement of the problem
Determining Your Need for Data Labeling for Machine Learning
Executing a test task
Determining Your Need for Data Labeling for Machine Learning
Estimation of timing and cost of marking
Determining Your Need for Data Labeling for Machine Learning
Writing instruction
Determining Your Need for Data Labeling for Machine Learning
Coordination
Determining Your Need for Data Labeling for Machine Learning
Result
Determining your need for data markup for machine learning