From the project archive
Pneumonia Detection with Deep Learning
Exploring chest X-ray classification with transfer learning
- My role
- ML Engineer
- Timeline
- 3 months
- Year
- 2022
- Format
- Project
- Reported precision
- 90%
- Model families
- VGG19 + ResNet
- Interface
- Desktop + voice
Context
This student project explored how transfer learning could classify chest X-ray images when labeled training data is limited.
Approach
Worked with a three-person team to compare VGG19 and ResNet convolutional neural networks using a chest X-ray dataset from Kaggle. Image preprocessing and parameter tuning supported the classification experiment.
The build
Connected the trained model to a PyQt desktop interface and added speech output so a user could receive the model's prediction visually and through audio.
Outcome
The project README reports 90% precision for the chest X-ray experiment. The student project brought together transfer learning, model comparison, and an accessible desktop interface.