Each idea is scoped for a two-semester project, with the tools you will need and a suggested plan. Adapt it with your supervisor.
Train a convolutional neural network to detect and measure cracks in concrete from photographs, to support faster building and bridge inspections.
Convert waste cooking oil to biodiesel and optimize catalyst amount, methanol ratio, temperature and time for yield and fuel quality.
Design a low-cost house using courtyards, shading, thermal mass and ventilation to cut indoor temperatures, and compare it with a typical local design.
Survey shopkeepers to find what drives or blocks the use of mobile wallets and QR payments, and recommend steps for banks and fintechs.
Detect shading, soiling, open-circuit and short-circuit faults in a small PV array from voltage, current and irradiance data.
Use five to ten years of published accounts to test whether models such as the Altman Z-score predict financial distress for PSX-listed firms.
Measure household or lab energy use in real time, show it on a web dashboard, and estimate which appliances are running from the load profile.
Monitor pH, dissolved solids and turbidity of irrigation or drinking water in real time and alert users when readings leave safe ranges.
Collect water from taps, filtration plants and tube wells around Multan, count and classify microplastic particles, and compare sources.
Forecast production from historical well data, comparing classical Arps decline-curve analysis with machine-learning models.
Survey residents about where and when they feel unsafe, map the results, and compare them with reported crime to suggest practical measures.
Predict how many operating cycles an engine has left from multivariate sensor data, using the public NASA C-MAPSS dataset, and compare classical and deep-learning models.
Test natural preservatives such as citric acid and plant extracts on mango pulp, measuring microbial growth, colour and taste over storage time.
Design, build and test an indirect solar dryer for local produce such as chillies, dates or mango slices, comparing drying rate and quality with open-sun drying.
Build a chatbot that answers prospective students’ questions about programmes, fees and deadlines using only the university’s official documents.
Create a small collection from offcuts and rejected fabric from Punjab textile mills, documenting sourcing, costing and the waste saved.
Build a low-cost vibration logger for a motor or pump, detect imbalance, misalignment and bearing faults from frequency spectra, and classify them with machine learning.