Python is the language most FYP groups already know, so the idea has to do the work. These 15 FYP ideas using Python go beyond chatbots and prediction models: automation, simulation, optimisation, computer vision and tools that people in Pakistani offices, schools and workshops would actually use. Each idea lists the Python libraries, where the data comes from and what you hand in.
None of these repeat the ideas in our main list of 70 FYP ideas for computer science students or our degree pages.
FYP ideas by degree: AI · software engineering · IT · cyber security · data science · web development · mobile apps · computer engineering (IoT) · business (BBA) · all 70 CS ideas
15 FYP ideas using Python
1. Excel report automation for offices (Easy)
Many offices still copy numbers between Excel sheets by hand every month. This tool reads the raw sheets, cleans them, builds formatted monthly reports with charts and emails them on a schedule.
- Libraries: pandas, openpyxl, matplotlib, and a scheduler such as cron or APScheduler
- Data: sample sheets from a real office or shop, used with permission and with names removed
- Deliverable: a desktop or web tool, and a before-and-after comparison of the time it saves
2. University notice aggregator with alerts (Easy)
Collects notices from several university and HEC web pages into one feed, tags them (admissions, exams, scholarships) and sends alerts by email or Telegram.
- Libraries: requests, BeautifulSoup, FastAPI, SQLite, python-telegram-bot
- Data: public notice pages; check each site’s robots.txt and terms, and keep your request rate low
- Deliverable: a searchable web feed and alert subscriptions
3. MCQ papers with OMR grading from phone photos (Medium)
Teachers build papers from a question bank and print shuffled versions with bubble sheets, then photograph the filled sheets and get marks, question-by-question analysis and a result sheet.
- Libraries: OpenCV, NumPy, pandas, and python-docx or ReportLab for printing
- Data: your own printed sheets, filled in by classmates
- Deliverable: a paper builder and a grader, with accuracy tested on 100 or more sheets
4. Code similarity checker for programming assignments (Medium)
Compares students’ submissions by structure, not just text, so renamed variables and reordered functions still match. It shows the matching parts side by side; the teacher makes the final call.
- Libraries: Python’s ast module, tree-sitter for other languages, difflib, Flask
- Data: anonymised past submissions, plus pairs you disguise on purpose
- Deliverable: a web tool and a test showing which disguises it catches
5. Urdu spell checker (Medium)
Finds misspelled Urdu words and suggests corrections, including common mix-ups between similar letters and words typed without a space between them.
- Libraries: rapidfuzz for edit distance, and an n-gram language model you build
- Data: a word list with frequencies from the Urdu Wikipedia dump, and a test set of real typing mistakes you collect
- Deliverable: a web editor or browser extension, with precision and recall on your test set
6. Duplicate records with Pakistani name spellings (Medium)
Finds the same person entered twice in a school, clinic or customer database, where names are spelled in many ways (Muhammad, Mohammad, M.) and phone numbers or fathers’ names differ slightly.
- Libraries: pandas, rapidfuzz, recordlinkage, scikit-learn for match scoring
- Data: synthetic records you generate with realistic spelling variants; never use real CNIC numbers
- Deliverable: a merge-review screen, with precision and recall
7. Degree plan checker (Easy)
Builds a prerequisite graph from a programme’s scheme of studies and plans a semester-by-semester path for students who failed a course or transferred, within credit-hour limits.
- Libraries: NetworkX, pandas, Streamlit
- Data: your department’s published scheme of studies
- Deliverable: a planner app, checked against real cases from your department office
8. Traffic signal timing simulation (Medium)
Models a busy junction in your city with its real lanes and signal timings, then compares fixed timings with timings that adapt to queue length.
- Libraries: SUMO with its TraCI Python API, pandas, matplotlib
- Data: OpenStreetMap for the road layout, and vehicle counts you take from video at peak hours
- Deliverable: a comparison of waiting times and queue lengths under each plan
9. Dengue spread simulation for a neighbourhood (Medium)
An agent-based model of people, mosquitoes and breeding sites that tests which measures slow an outbreak most: clearing standing water, fogging or awareness drives. It compares measures; it does not forecast real cases.
- Libraries: Mesa, NumPy, matplotlib
- Data: model parameters from published studies, cited in your report
- Deliverable: an interactive simulation and a sensitivity analysis
10. Ludo AI with Monte Carlo tree search (Medium)
A Ludo game where the computer player chooses moves with Monte Carlo tree search, measured against simple rule-based players.
- Libraries: Pygame, NumPy
- Data: none needed; you generate games by self-play
- Deliverable: a playable game and win rates over thousands of simulated games
11. Ball tracking for street cricket (Hard)
Tracks the ball in phone video shot from behind the bowler, draws its path and predicts whether it would have hit the stumps.
- Libraries: OpenCV, NumPy, SciPy for curve fitting
- Data: videos you record at your own ground, with known outcomes
- Deliverable: a video overlay of the ball’s path, and accuracy on recorded deliveries
12. Signature verification for cheques and forms (Hard)
Checks whether a signature matches the reference signatures on file and gives a confidence score for a person to review. It supports a decision; it does not make one.
- Libraries: PyTorch, OpenCV
- Data: public signature datasets such as CEDAR, plus signatures you collect with consent
- Deliverable: a verification tool with false accept and false reject rates
13. Cutting plan optimiser for workshops (Medium)
Furniture and steel workshops often cut boards and rods by eye and waste material. This tool takes the pieces an order needs and the stock sizes available, and plans the cuts with the least waste.
- Libraries: Google OR-Tools or PuLP, and matplotlib for cut diagrams
- Data: real orders from a local workshop
- Deliverable: printable cut diagrams, and the material saved against the workshop’s usual method
14. Radio ad airing checker (Hard)
Advertisers pay for ads on FM radio but cannot easily check that they aired. The system records a station and finds each ad by its audio fingerprint, with the time it played.
- Libraries: librosa, NumPy, SciPy, with a landmark-hashing fingerprint method
- Data: recordings of a station for research use, and your own ad clips
- Deliverable: an airing report, with detection accuracy
15. Wi-Fi coverage heatmap for a building (Medium)
Collects Wi-Fi signal strength at points around a floor, draws a coverage heatmap on the floor plan and suggests where to move or add access points.
- Libraries: pandas, SciPy for interpolation, matplotlib, and system commands or pywifi to read signal strength
- Data: your own measurements on a floor plan of your department
- Deliverable: before-and-after heatmaps after the changes you suggest
Short on time? We can build it for you.
Ezitech builds final year projects for students across Pakistan — Rs 25,000, delivered in 3 days. Send your idea and your supervisor’s requirements on WhatsApp and we will tell you straight away whether we can do it.
Frequently asked questions
Is a Python FYP enough, or do we need another language?
Python alone is enough for most of these ideas. Add JavaScript only if you need a richer web front end; Streamlit or Flask templates are fine for many projects.
Django, Flask or FastAPI?
Use Django when you need an admin panel, user accounts and many database tables. Use FastAPI for an API that a mobile app or front end calls. Use Flask for a small tool with a few pages.
Can we use data we scrape from websites?
Only if the site’s terms and robots.txt allow it. Keep your request rate low, store only what you need, and never collect personal data without consent.
Next steps
Still choosing? Read how to choose a good FYP. Picked an idea? Write it up with our FYP proposal format and the SRS template for FYP, find a name in 60 FYP title examples, or filter more ideas with the FYP idea generator.
Want to use Python on real projects before the build starts? See the Ezitech machine learning internship, the data science internship and the Python developer roadmap.