← All work
2021
Deep Learning, Computer Vision
Skin Cancer Prediction iOS AppBSc Thesis
An iOS app that classifies skin lesions with a VGG16 pre-trained CNN, served from a Python/Flask backend to a Swift frontend.
In numbers
- 82%
- classification accuracy
What I did
- 1.
Fine-tuned a VGG16 pre-trained convolutional neural network to classify skin lesions from dermoscopic images, reaching 82% accuracy.
- 2.
Served predictions from a Python/Flask API to a native Swift iOS app that captures or uploads images for real-time analysis.
Built with
Python · TensorFlow · VGG16 · Flask · Swift
Next project
DevSweep →