12 Aug Comparative evaluation of Histogram of Oriented Gradients (HOG) and a Convolutional Neural Network trained
Q1: Summarize the paper titled “An Introduction to Convolutional Neural Networks” by Keiron O’Shea and Ryan Nash (https://arxiv.org/pdf/1511.08458.pdf ) in your own words in two pages.(Points: 50)
Q2: Comparative evaluation of Histogram of Oriented Gradients (HOG) and a Convolutional Neural Network trained from the scratch (which means designing your own CNN) for gender classification task on a subset of CelebA dataset. Provide detailed information on the steps involved and record your observations ((a) which model performed better and by how much).
Deliverables:
– A well-documented report containing steps involved, your model summary, information on hyperparameters (optimizer, epochs, loss function) used for training the model, and test accuracy. Record the observations on comparative evaluation of HOG and the CNN. The code should be uploaded as well.
– One- or two-page write-up of the aforementioned paper (“An Introduction to Convolutional Neural Networks”) as a pdf.
