Research Projects

High Impact Skill Development Program for GB Youth

The demand for AI and Blockchain experts is increasing globally and having training in these fields will make Gilgit-Baltistan an asset to companies and organizations around the world. To promote the ecosystem of research and employable skills in Gilgit-Baltistan, NUST in collaboration with AI Lounge (pvt) Ltd., is providing six-months intensive training to students and learners who have a strong aptitude to learn and grow in Data Science, Artificial Intelligence, and Blockchain.

• PI: Prof. Dr. Faisal Shafait

Early Detection of Forest Health Decline using Remote Sensing

The world's forests are under significant danger due to climate change and global warming. There are two major natural disasters that can rapidly impact the large forest areas. First, forest fires, and second, forest diseases. We aim to explore the deep learning based semantic segmentation topologies for early detection of the infections in the trees using remote sensing imagery.

• German PI: Prof. Dr. Norbert Wehn
• Pakistani PI: Prof. Dr. Faisal Shafait

Cotton Pest Prevalence Estimation System for Early Warning using UAV & AIoT (CoPEST)

Early prediction of pest attacks can be very helpful in improving productivity in agriculture. To this end, we propose to develop an AI-based IoT-enabled system for early detection of cotton pests using the data of environmental parameters and multispectral images acquired from multiple sensors, which help to alert the farmer of taking prevention measures hence producing high crop yield.

• German PI: Prof. Dr. Karsten Berns
• Pakistani PI: Dr. Zuhair Zafar

Forest Monitoring and Change Detection using Remote Sensing Data

To develop a system that is able to quantitatively monitor forests of Pakistan by deploying a small low-cost UAV to acquire temporal images and possibly fuse it with available satellite imagery to enable thorough assessments of local forest conditions e.g., degradation and deforestation caused by theft, urbanization or natural factors such as arid climate, heavy dependence on irrigation water, etc. by employing state-of-the-art machine learning techniques.

Read More

• German PI: Prof. Dr. Norbert Wehn
• Pakistani PI: Dr. M. Shahzad

Crop Monitoring via Remote Sensing from Earth Observation Satellites to UAV

To generate a crop monitoring system that enables the user to assess the crop’s phenological growth stages and based on extracted crop parameters, predict the crop yield productions. The developed system is intended to assist in filling gaps in policy makers information that is essential in ensuring sustainable management of crops and forecast yield predictions.

• German PI: Prof. Dr. Xiao Xiang Zhu
• Pakistani PI: Dr. M. Shahzad

IoT Enabled Surface Water Pollution Detection for Predictive Healthcare

To develop an Internet of Things (IoT) based system to monitor the quality of water in Pakistan, thus, providing a proper and near real time assessment of the water to the community in a ubiquitous fashion to overcome the limitations of conventional costly and inefficient methods of water quality monitoring. The purpose is to improve the performance as compared to the state-of-the-art.

• German PI: Prof. Dr. Andreas Dengel
• Pakistani PI: Dr. Rafia Mumtaz

Water resource estimation by Satellite Based Image analysis (WASABI)

In order to manage Pakistan’s water resources, it is necessary to enforce an even distribution of water throughout the year. WASABI is designed to predict flooding beforehand, supervise the volume of water bodies (including rivers, lakes, and reservoirs like dams, etc.), and predict the ratio at which the water level is declining or even increasing during some parts of the year, yielding accurate estimates about the future supply of water.

Read More

• German PI: Prof. Dr. Adrian Ulges
• Pakistani PI: Dr. M. Imran Malik

Wearable EEG for prediagnostic screening of Mental Diseases in Rural Areas

The aim is to develop a low cost, wearable device that can record several physiological parameters simultaneously, and which can communicate with a smartphone. The most relevant parameters to be handled are EEG, heart rate, and human body motion. The system will serve to reduce the workload of the neurologist who would not need to spend time on reading perfectly normal recordings.

• German PI: Prof. Dr. Didier Stricker
• Pakistani PI: Dr. Awais M. Kamboh

The Fish Biodiversity Estimation by Low-Cost Non-Destructive Video Based Sampling (2016)

To employ underwater computer vision technology for an automated surveying of fish species. The project will form the basis for a non-invasive, low-cost surveying of Pakistan’s fish population, and benefit the preservation of endangered species as well as a more sustainable economic exploitation of Pakistan’s resources.

• German PI: Dr. Adrian Ulges
• Pakistani PI: Prof. Dr. Faisal Shafait

AI Driven Decision Support System for Pakistan Judiciary

This project focuses on basic aspect of automation in the overall picture of judicial automation. It creates a Knowledge Base and after that Data Anonymization, Named Entity Recognition, Similar Case Retrieval and Verdict Recommendation is performed.

• PI: Prof. Dr. Faisal Shafait

Media Monitoring

The aim is to develop a robust Media Analytics platform for mass media using intelligent monitoring systems to provide valuable analytics as a service to keep all the stakeholders of the media industry well informed. This can help key media stakeholders to streamline processes such as media monitoring & measurement, quality assurance, compliance, media content research and identification of program clutters, which in return will boost revenue, and increase customer satisfaction.

• PI: Prof. Dr. Faisal Shafait

Customized Irrigation and Climate Advisory Service through Citizen Science

This project in collaborate with the Pakistan Council of Research in Water Resources (PCRWR) focuses to evolve an already existing Irrigation Advisory Service (IAS) to Irrigation and Climate Advisory Service (ICAS). This evolution will lead to accurate weather forecasting by taking into account satellite data, Citizen Science Data, and Automatic Weather Station (AWS) data.

• PI: Prof. Dr. Faisal Shafait

An Automatic Real-Time Vehicle Detection, Identification and Registration Plate Recognition System (VSURV)-(2017)

To develop an automatic real-time vehicle registration number detection and recognition embedded device using machine learning and computer vision based techniques, it will automatically detect a vehicle and its registration number from live feed of the already installed IR video cameras.

• PI: Prof. Dr. Faisal Shafait

Reshaping Online Consumer Experience with Data Driven Personalized Chatbots-(2017)

To develop a bot that could augment the intelligence of a user in order to retrieve such content that could satisfy the potential needs of the user. The bot will be able to handle user queries in the real estate domain for Pakistani target market. The project also aims to develop an extensive database of the housing and property inventory existing in Pakistan.

• PI: Prof. Dr. Faisal Shafait

Crop Health Monitoring and Early warning system using IoT enabled precision agriculture (2019)

To implement an extensible IoT-based crop health monitoring system which will help farmers to overcome conventional techniques of farming i.e. minimize manual and inefficient methods of crop monitoring and to use technology-based solutions for informed decisions related to crop under stress instead of their intuition; consequently this will improve crop yield and have a positive impact on Pakistan’s GDP which is based on farming.

• PI: Dr. Rafia Mumtaz

Non-destructive approach for detecting Marine life in Fresh Waters of Pakistan - (2016)

The goal of the proposed project is to employ underwater computer vision technology for an automated surveying of fish species by providing state-of-the-art automatic, non-destructive, cost-effective, and reliable solution. It will be able to detect and classify the available species of fresh water fish and a database of the fish species will also be maintained.

• PI: Dr. M. Imran Malik

iDoc: A portable, cost-effective diagnostic screening device for early dengue fever detection (2017)

The aim of this project is to develop a cost effective and portable embedded remote medical checkup device that provides an accurate and rapid solution for an early detection and diagnosis of Dengue Fever/ Dengue Hemorrhagic Fever/ Dengue Shock Syndrome. The device can also be used by the Lady Health Workers (LHW) and dispensers at the remote locations where the medical facilities are non-existent.

Read More

• PI: Prof. Dr. Faisal Shafait

Design and Development of Condition Monitoring Testbed based on Smart solar Microgrid

This project aims to develop the condition monitoring system as well as the intelligence behind the “smartness” of a smart grid. For this purpose, the model for test micro grid was developed which was able to give energy demand estimate in correlation with the weather conditions. Such technologies will be vital to achieve a smart and optimized solar micro grid for a marginalized community in Pakistan which can not only improve their quality of life, but also enhance their abilities and their contribution towards the country’s economy.

• PI: Syed Muhammad Raza Kazmi

Intelligent Data Extraction from Mortgage Closing Disclosure Forms using OCR and Natural Language Processing - Acrux Technologies

To develop a machine learning software that can extract all the relevant data from a sample form (PDF) under the following assumptions;
● There will be only one type of form that will be processed.
● Up to three variations of this form will be considered.
● Approx 100+ samples will be provided by client in readable PDF format.

• PI: Prof. Dr. Faisal Shafait

Document Image Understanding and Editing using Tesseract OCR - DocIT Solutions

This research work aimed at automating paper-based mailbox of large companies by developing an automated document understanding system. To make this process efficient, the system was developed incrementally in order to extract the information from scanned documents.

• PI: Prof. Dr. Faisal Shafait

Cursive Urdu Script Recognition using Deep Learning- 2015 NVIDIA

The project aimed to perform OCR on an extensive dataset of printed and handwritten Urdu documents. The computational power of the GPU, managed by TensorFlow, enabled us to perform extensive research on training uni or multi-directional LSTM in order to successfully develop an OCR of complex-script documents.

• PI: Prof. Dr. Faisal Shafait

Machine Learning For Combating COVID-19

The spread of coronavirus in over 200 countries within a span of a few months necessitates the development of forecasting models that can assist governments around the globe in keeping an eye on their country’s situation forecast so that preventive measures can be timely taken. Since the epidemic spread is still in the early stages in some countries, forecasting models harnessing data from countries already hit by the epidemic will provide a realistic and timely intimation of risks of potential outbreaks in these countries. The proposed open-source data...

Read More

• PI: Prof. Dr. Faisal Shafait