broutonlab.com Open in urlscan Pro
104.248.230.237  Public Scan

Form analysis
0 forms found in the DOM

Text Content

Our Team
Success Stories

Resources

 * Research
 * AI in Action
 * Solutions
 * Projects
 * Blog
 * Our Webinars

Solutions Projects AI in Action Blog Research Our Webinars
Get Started



BUILD YOUR FRACTIONAL DATA SCIENCE TEAM NOW

Hire and manage top data scientists and engineers to get your projects done
efficiently

#DataEngineers
#MachineLearning
#DataScientists
#BackendDevelopers
#ArtificialIntelligence
Get Started





HOW WE HELP STARTUPS

 * We work as an extension of your team.
   
   

 * We bring top talent and expertise to your business to handle complex data
   science and AI projects.
   
   

 * Our model is flexible, scalable and offers total transparency.
   
   

 * We help start-ups, investors, and programmers solve real business problems,
   reduce time to market, and optimize processes.






GET MEASURABLE VALUE AT EACH STEP

 * We work by quick, week-long sprints.
   
   

 * Results of each sprint are measurable and you do not have to wait extra time
   to test them.
   
   

 * If you want, you can keep iterating with us to improve the accuracy to the
   level you need.






SUCCESS STORIES


AUTOMATED SPORTS FAN VIDEO GENERATION WITH FACIAL RECOGNITION


AI FOR RESUME PARSING AND JOB MATCHING


EXPLICIT CONTENT DETECTION MODEL


OCR CHEQUE SCANNER FOR A BANKING APP

All success stories



GO FROM IDEA TO PROTOTYPE


30 HOUR PROJECT EXAMPLE

MRI tumor segmentation

PROBLEM:

Radiologists read many medical images each day and occasionally misread them.
Your team would like to provide doctors and radiologists help with a secondary
automated MRI analysis. Your goal is to create a model that can identify tumors
on MRIs quickly and precisely to provide improved diagnostic results.

INPUTS:

We will need a dataset of MRIs with tumor regions labeled by doctors.

OUTCOME:

In 30 hours, we can create a model with significant accuracy that can find and
identify tumors in medical imagery. The model can then be fine tuned over time
and with additional data to achieve your desired level of accuracy.





20 HOUR PROJECT EXAMPLE

Maximize Customer Lifetime Value For A Mobile App

PROBLEM:

You have several subscription options for your new mobile app and you would like
to recommend the best option for each customer to maximize potential revenue and
customer lifetime values. You have been running a/b tests and collecting
customer data to help understand customer preferences.

INPUTS:

We will need the customer data obtained during your tests for different
subscription options.

OUTCOME:

In 20 hours, we can provide critical information about your customers' lifetime
value and churn rates, and create a model of monthly and annual recurring
revenue.





ONE-CLICK INTEGRATION WITH YOUR SOFTWARE

Save your developers time by working with us

 * We will prepare a production - ready docker container with a REST API or an
   SDK to embed in your mobile app.
   
   

 * Your developers will easily deploy our model and integrate it with your app.





PROBLEMS SOLVED WITH AI


YODA VOICE GENERATOR

SOLUTION

We propose a novel approach to the end-to-end text-to-speech (TTS) generation
with some custom voice style.

AUDIO GENERATION WITH CONSTANT VOICE

For this step we use a pretrained FastSpeech model for the mel-spectrogram
generation and a pretrained MelGan model for translation of this mel-spectrogram
to voice audio.

AUDIO TRANSFERRING TO TARGET VOICE STYLE.

We perform audio style transfer based on the MelGan architecture. We collect
target voice audio samples and train the transfer network to change the voice
style of the recording to desired voice style. Each voice requires a unique
transfer training.




LICENSE PLATE NUMBER RECOGNITION

We propose a robust ALPR system implemented in the form of a pipeline solving
the problem.

The pipeline comprises:

 * vehicle detection.

 * vehicle plate number location recognition.

 * plate number optical characters recognition.

To improve the quality of the pipeline, an object tracker has been embedded into
the system. State-of-the-art neural network approaches has been used for all
parts of our pipeline. Moreover, the proposed solution outperforms other methods
on the OpenALPR benchmark.




YODA VOICE GENERATOR

SOLUTION

We propose a novel approach to the end-to-end text-to-speech (TTS) generation
with some custom voice style.

AUDIO GENERATION WITH CONSTANT VOICE

For this step we use a pretrained FastSpeech model for the mel-spectrogram
generation and a pretrained MelGan model for translation of this mel-spectrogram
to voice audio.

AUDIO TRANSFERRING TO TARGET VOICE STYLE.

We perform audio style transfer based on the MelGan architecture. We collect
target voice audio samples and train the transfer network to change the voice
style of the recording to desired voice style. Each voice requires a unique
transfer training.




LICENSE PLATE NUMBER RECOGNITION

We propose a robust ALPR system implemented in the form of a pipeline solving
the problem.

The pipeline comprises:

 * vehicle detection.

 * vehicle plate number location recognition.

 * plate number optical characters recognition.

To improve the quality of the pipeline, an object tracker has been embedded into
the system. State-of-the-art neural network approaches has been used for all
parts of our pipeline. Moreover, the proposed solution outperforms other methods
on the OpenALPR benchmark.



01

/

02





CUSTOMER REVIEWS


CLARENCE LEE

Digital Marketing Professor at Cornell



Michael's developers are amazingly capable and precise -- they helped me
translate my code for a hierarchical Bayesian model MCMC estimation routine from
R/C++ into CUDA, which helped me speed up the computation by 2 orders of
magnitude due to the fact now I was able to parallelize the code onto high-power
nVidia GPU's.


HUNTER HAWLEY

CEO and Founder of Blueprint Stats, Indiana



BroutonLab team helped me automate our workflow with their AI development
skills, and it sped up the rate at which we process our customers’ requests.
Getting started with them was easy, and the results were outstanding!


ADRIAN MIHAI

CTO and Co-Founder of Opening, Dublin. Acquired by iCIMS



I am more than happy to have interacted with Mikhail. Besides his amazing
personality he has lots of patience and does his best at managing timelines.
It's with utmost delight I'll recommend him now - an extremely trust-worthy and
talented engineer with a treasure trove of experience and ethic. 5/5




GET STARTED FAST

We start working on your project without long-term commitments or hefty bills:

 * Send us your brief project description
   
   

 * Outline the goals of your project

We will get back to you with proposal shortly.

$60

USD/hr

from 10

hour/week
Get Started




SUCCESS STORIES

Video-Based Face Recognition for a Sports App. NLP for Job Matching and Talent
Intelligence Real-Time Image Recognition for an Android App OCR Cheque Scanner
for a Banking App

SOLUTIONS

License Plate Recognition Facial Recognition Automated Monitoring of Solar
Panels AI-Powered Synthetic Data Generation Data Extraction and Document Parsing
Software

SEE AI IN ACTION

Masks License Plate / Vehicle Recognition Fashion Tagging Demo Food Recognition
Parking Monitoring System Legal Documents Analysis with NLP

RESOURCES

Research Blog Success Stories Our Team Write For Us Partnerships Projects
Contact

© 2017-2021. BroutonLab.

Privacy policy


THANK YOU FOR YOUR APPLICATION

We will reply to you as soon as possible

BroutonLab.

Your message was successfully sent!

Ok