leorazhang Goto Github PK

Intro: Future wizards and witches, are you curious about which Hogwarts House you belong to in the magic world? Let's find out the secret with my code-made Sorting Hat! An ##image classification model## is trained to classify ordinary people into 4 Hogwarts Houses based on the face features of Harry Potter characters. This can be applied to entertainment use, like an interactive game in the Universal Studio to attract tourists or an online testing game to draw natural traffic to the official website. The problem is that the number of characters in the Harry Potter series is really limited, so the datasets I made for the training and validation processes are pretty small and unbalanced. Besides, the accuracy of validation data is low because there might not be reliable patterns for the classification. As we all know, the Sorting Hat doesn't do its work according to people's appearance(while the Weasley family with red hair might be an exception). However, this model is mainly for exploration and fun, so let's put our classmates' faces in the model to have a try!