Capstone
Information about Capstone Design projects.
Capstone Projects
✅ General Deliverables (for every project)
- Proposal Document (Week 3): Problem, objectives, timeline.
- Midterm Presentation (Week 7): Progress update.
- Final Deliverables (Week 12):
- Trained ML model
- Working demo (Streamlit app or Arduino)
- Final report (methods, results, discussion)
- Slides + oral presentation
Project Ideas
1. Hybrid Solar + Grid Power System (Bangladesh Focus)
- Theme: Renewable Energy + Simulation + Arduino + ML
- Dataset/Tools:
- HOMER / MATLAB (simulation),
- Arduino with small solar cell,
- weather/solar datasets (Kaggle/NREL).
- Deliverables:
- HOMER design of solar + grid backup system.
- Arduino demo with LED load switching between “solar” and “grid.”
- Simple ML model predicting daily solar power from weather data.
2. Football Match Outcome Predictor
- Theme: Sports Analytics + ML
- Dataset/Tools:
- Kaggle “European Football Dataset,”
- FBref stats (Barcelona).
- Deliverables:
- Exploratory data analysis of team/player stats.
- ML model to predict win/draw/loss given past performance.
- Streamlit demo: input two teams → see win probability.
3. Solar Panel Efficiency Predictor
- Theme: Energy + IoT + ML
- Dataset/Tools:
- Arduino + small solar panel (optional),
- or Kaggle/NREL open solar datasets.
- Deliverables:
- Data collection (voltage/current or weather vs. solar output).
- ML regression model predicting solar efficiency.
- Tangible demo: Arduino or Streamlit graph of real vs. predicted performance.
4. Crop Disease Detection (Image Classification)
- Theme: Computer Vision + Agriculture
- Dataset/Tools:
- Kaggle “PlantVillage” (healthy vs. diseased leaves).
- CNN in TensorFlow/PyTorch.
- Deliverables:
- Preprocessing of images + CNN training.
- Model achieving >80% accuracy.
- Streamlit app: upload leaf → model predicts disease type.
5. Smart Waste Sorting System
- Theme: Computer Vision + Sustainability
- Dataset/Tools:
- TrashNet dataset (glass, plastic, paper, metal).
- CNN in PyTorch/TensorFlow.
- Deliverables:
- Train CNN to classify waste items by category.
- Accuracy evaluation + confusion matrix.
- Tangible demo: Arduino with 4 LEDs (red=plastic, green=glass, etc.) lights up based on predicted category.
6. Student Performance Predictor
- Theme: Education + Data Science
- Dataset/Tools:
- UCI “Student Performance Data,”
- Kaggle exam success datasets.
- Deliverables:
- EDA of student performance factors (hours studied, GPA, attendance).
- Regression/classification model to predict final grade.
- Streamlit demo: enter features → get prediction of success.
7. Air Quality Monitoring and Prediction
- Theme: IoT + Environment + ML
- Dataset/Tools:
- Arduino air quality sensor (PM2.5)
- OR Kaggle “Air Quality” datasets.
- Deliverables:
- Collect or download AQI data.
- ML regression model to predict AQI from weather/traffic.
- Arduino or Streamlit demo: display “Good / Moderate / Unhealthy” classification.
