MNIST
Created a computer vision model to recognize handwritten digits with 99% accuracy. Model ranked in the top 36% globally of all Kaggle users. Used a Convolutional Neural Network implemented in TensorFlow.
What I learned
- This was my first complete solo computer vision project.
- I experimented with many CNN types, including LeNets, AlexNets, and GoogleNets.
- I also learned how to create a complete learning pipeline, including simple image processing.