MIL

iDog is a product based on our own OCR technology. The MIL team has implemented a product that helps companies in the paperwork process.

POSSIBILITIES OF OUR SOLUTION
1
FAST LEARNING TO RECOGNIZE NEW TYPES OF DOCUMENTS AND NEW LANGUAGES
2
POSSIBILITY OF TESTING DIFFERENT OCR MODELS WITH ASSESSMENT OF THE BUSINESS PROBLEM SOLUTION ACCURACY
3
OUR SOLUTION ALLOWS TO WORK WITH SPECIFIC CASES - RECOGNITION OF TABLES, SIGNATURES AND SEALS, PERSONAL DOCUMENTS
The iDog solution is based on the use of neural network models for character recognition and text detection. It outperforms open-source analogues both in quality and in the speed of work.
WE HELP TO SOLVE OUR CUSTOMER'S TASKS
Determine the optimal OCR solution
The Comparison module of the iDog library allows you to evaluate different OCR solution providers and select the exact solution that provides the best quality for your business task.
Retrain in new languages
Our library can be quickly retrained to recognize new characters or languages since we have created technology that learns from synthetic sampling with high quality.
Retrain in new documents
A flexible system for presenting documents in the format of our system allows you to quickly annoate the structure of a document and, in the future, automatically retrieve information from documents of a similar structure.
Recognition of personal documents in a stream
The iDog library module allows you to quickly create a document template with a fixed structure and recognize a large stream of such documents with high quality and speed.
EXAMPLES OF IMPLEMENTED WORKS
WHAT TECHNOLOGIES ARE USED INSIDE
What we use to code
Main programming language: Python
Backend: Python Flask, Django
Frontend: React, vue.js
Working with data: PostgresSQL, MongoDB
Distributed computing: Spark, Hadoop
Notebooks: Colab, Jupyter, H2O;
Model packaging and data flow control: Kubernetes, Docker, Airflow;
Profile frameworks
Neural network frameworks: PyTorch (prototyping), TensorFlow (production), TensorFlow lite (import to devices);
Working with data: pandas, numpy;
Profile libraries: BigARTM, TopicNet, gensim, nltk, DeepPavlov, SpaCy, OpenCV, scipy, etc;
Experiment management: Wandb, MLFlow, Tensorboard;
Code Style and solution delivery
  • Styling your code as python scripts using PEP 8
  • High level of code readability
Delivery options:
  • Docker + REST API;
  • Web-service + Frontend;
  • Python-scripts;
  • Python libraries.
OUR PRICE POLICY
FINISHED MODULES
— Text Detection and OCR
Detection and segmentation of tables
— Comparison of solutions
500к+
Module
ADAPTATION ON REQUEST
— Text Detection and OCR
New type of documents
— Retraining a new language
3к+
per specialist's hour
HOW WE WILL WORK TOGETHER
1
Problem statement and business context
Defining a set of tasks, the solution of which is relevant for business.
Discussing the wishes and limitations that the company has.
2
NDA and cooperation agreement
Signing a non-disclosure agreement.
Signing a framework cooperation agreement.
3
Assessment and project objectives
Defining the business goals in solving the problem and the indicators of the project's success.
Preparing a technical task and a commercial proposal for the stages of solving the problem.
4
Contract
Agreeing on the terms of the contract and sign the contract on both sides.
5
Task data and questions
Preparing a data slice to solve the problem.
Discussing the questions that arose during the analysis of the problem.
6
Solution and delivery
Carrying out several iterations of solving the problem.
Delivering ready-made results according to the agreed project roadmap
7
Delivery and acceptance of the project
Order a project
To start the project, we need to talk. Just fill in the fields below and we will contact you.
Promo code