Project information

• Category: Artificial Intelligence, Vision
• Project Tools: Puthon, Google Colab
• Project date: May, 2022
• Project URL: GitHub Repository

Project description

In this project, we want to use different neural networks (specifically fully connected networks) to build a model that takes photos as input and recognizes the category of each photo. We implement the feed-forward, backpropagation, and other methods from scratch. Our dataset is CIFAR-10, which is a smaller version of the CIFAR-100 dataset, which is limited to images from 10 different categories.

To reduce the computational volume, we only use the images of the first 4 classes of this dataset. Consequently, our output layer will contain 4 neurons. The neuron with the highest activation value is selected as the category detected by our model. For this neural network, we consider two hidden layers, each of which has 16 neurons.