IDEAS
Technology Innovation Hub
Indian Statistical Institute, Kolkata
Our Products
Build, test, and deploy data-driven applications quickly with our in-house platforms
Upatto is a robust, multi-user web-based enterprise application that supports the entire lifecycle of data engineering and ML model building. It offers flexible deployment on premise or in the cloud , providing a scalable alternative to platforms like Databricks.
Upatto
XInsight: A lightweight, single-user desktop tool tailored for rapid PoC development and insightful visualisation of numerical data. It serves as a powerful no-code substitute for commercial products like Power BI or Tableau.
XInsight
AudioFill is an intelligent voice-driven data automation system that transforms spoken input into structured digital forms in real time. By leveraging advanced audio processing and AI-powered parsing, it seamlessly converts natural speech into accurate, organized data—eliminating manual entry and reducing errors. Designed for modern research, enterprise, and academic environments, AudioFill delivers a fast, intuitive, and efficient approach to form completion, enhancing productivity while maintaining a clean, user-centric experience.
AudioFill - Convert spoken input into structured digital forms in real time
Consultancy Services
Real-world impact, delivered cost-effectively
Statistical Consulting
-
Statistical Data Analysis & Insights
-
SQC, OR & Process Control
-
Experiment, Survey & Observational Design
-
Domain-specific Study / Analysis
Data Science & AI/ML
-
Data Analytics, Data Engineering & Data Management
-
Machine Learning Model Development
-
Computer Vision Solutions
-
Generative AI & NLP · Agentic AI
Application Development
-
Web Application Development
-
Mobile Application Development
-
Software Engineering
-
LLM integration
-
Analytics Dashboard
-
Agentic AI
Assignment:
Municipal tax assessment using drone imagery and geofencing data.
IDEAS ISI is collaborating with the Indore–Bhilai Innovation & Technology Foundation (IBITF), IIT Bhilai, on a strategic project for the Bhilai Municipal Corporation. The initiative focuses on large-scale geospatial and historical data analysis using computer vision and 3D LiDAR technologies. Key components include cleaning and annotating existing tax records, developing a new geospatial database, building computational pipelines to detect changes in historical imagery, and extracting property attributes such as area and number of storeys from LiDAR data. The six-month project aims to enable accurate, data-driven urban assessment and improved tax governance.
Assignment:
Reduce Manufacturing Costs and Improve Quality of Cartridge Cases at Metal and Steel Factory, Ishapore
This project aims at reducing manufacturing costs through application of SPC, Measurement System Analysis and related techniques leading to reduction of inspection and RFR. By focusing on eliminating waste, reducing process variability, and actively engaging the workforce in continuous improvement, this project will enable MSF to reduce cost of Cartridge Cases to a large extent.
Assignment:
PoC development: Survey Anomaly Detection using Statistical and Machine Learning based Approach
The National Statistical Office (NSO) under MoSPI is responsible for collecting, processing, and disseminating statistical data to support evidence-based policymaking. As part of its digital transformation, NSO uses CAPI tools and the eSigma platform for survey data collection and management. While existing validation checks ensure basic consistency, they are rule-based and unable to detect complex anomalies such as enumerator bias, temporal shifts, or inconsistencies with historical data. Supervisors currently lack tools to leverage past datasets for deeper validation. To address this gap, a proof-of-concept has been developed by IDEAS that uses probabilistic and machine learning models to enable advanced, context-aware validation of survey data in both online and offline modes.
Assignment:
Software development for CPI Survey: Govt of Tripura
IDEAS TIH has developed a mobile app for Govt of Tripura, Directorate of Economics and Statistics, for smooth collection, reporting and analysis of consumer price index data. The app is expected to result in susbtantial effort saving and cost reduction for the government
Assignment:
FASAL 2.0 - Development of National Crop Yield Forecasting platform based on statistical and machine lerning models on multi-source data (for Ministry of Agriculture and Farmers Welfare)
A platform is being developed for the Ministry of Agriculture and Farmers Welfare leveraging an advanced ensemble of statistical, econometric, and deep learning models to provide high-precision yield predictions for 10 of India’s most economically critical crops, driving data-informed stability in the agricultural sector. This is a 3 year project assignment.
Case Studies
See how ideas translate into industrial impact
Defect Detection in Manufacturing Industry
Click on the image to enlarge
Enhance quality assurance with our AI-driven industrial defect detection solution, designed to deliver precision, speed, and scalability across manufacturing operations. Powered by advanced deep learning and computer vision technologies, the platform identifies surface anomalies, structural inconsistencies, and production defects in real time with high accuracy. It minimizes reliance on manual inspection, reducing human error and operational costs while ensuring consistent quality standards. Built for seamless integration with existing production lines, the solution supports diverse materials and defect types, adapting to dynamic industrial environments. With actionable insights and automated monitoring, organizations can improve yield, reduce downtime, and accelerate decision-making — driving operational excellence and competitive advantage in modern manufacturing.
Unlocking Crowd Intelligence with Thermal & Infrared Video Analytics
Click on the image to enlarge
Our cyber-physical platform leverages advanced thermal and infrared computer vision to deliver actionable insights into crowd dynamics, operating seamlessly even in challenging environments and lighting conditions. System Highlights:
1. Real-Time Anomaly Detection: Deep learning algorithms automatically identify critical events, behavioral patterns, and density variations.
2. Scalable Architecture: Efficiently processes complex sensor data through a robust cloud pipeline for rapid decision-making.
3. Enhanced Public Safety: Reduces manual observation and empowers predictive, data-driven resource allocation.
Partner with us to transform your crowd monitoring into intelligent, predictive management.
Monitoring Blast Furnace Parameters using Machine Learning Techniques
Click on the image to enlarge
Real time monitoring Top gas velocity, Burden shift, Burden leakage of Industrial blast furnaces using video analytics.
Estimation of Ferrite, Martensite, and Bainite in Steel using Machine Learning
Click on the image to enlarge
Work undertaken for a leading steel manufacturer in India, the technology developed includes application of computer vision and machine learning techniques to accurately estimate these microstructures that largely impact the steel manufacturing process.
Development of AI analytical models to detect potential aberrations in phono-cardiogram signals – A project for Cardiological Society of India
Click on the image to enlarge
IDEAS is developing a smart health monitoring system for screening of children for congenital heart diseases which will result in early detection of cardiac murmur to prevent developing Rheumatoid Heart Disease in adults. Phono Cariogram Signals from hundreds of children have been analysed using deep neural network to develop a prediction model for children heart murmur. The project is currently in field testing phase.