Details: this model will be built using traditional machine learning, with an effort to identify whether there exists a certain fruit in a given image (object detection). This project will be written using a tensor library called CuPy, which is a library that mirrors the NumPy API but with GPU acceleration. Expect to write backpropagation, gradient descent, MLPs, and CNNs from scratch.
Team size: 3
Progression:
- learning how autograd works
- implementing an autograd library
- writing a simple MLP
- creating a simple Linear layer
- creating Sequential, Conv2D, MaxPool2D, etc.
Technologies: CuPy/NumPy, Python