monkeykingg / anime-face-detection-and-recognition

CSC420 Fall 2019 Project

Anime Face Detection and Recognition

CSC420 Fall 2019 Project

Team Members:

Zhicheng Yan
Zhihong Wang

Ideas:

Anime becomes more and more popular. Sometimes, people will find interesting anime characters from images without recognizing their information (e.g. the name of the character). We intend to help anime enthusiasts to find the information of characters from their anime images. For our project, we are planning to perform the anime face detection on images. It will draw a rectangular box on the face of the input anime image. After the detection, we will perform the recognition, which means our project will tell the character's information from the anime face by classification.

Tasks:

1. Get Datasets

Find anime image datasets or collect anime images and create own datasets, which are large enough and contain recognizable anime faces. Datasets with labels should be preferable. We may resize them if needed.

2. Anime Face Detection

Detect and locate anime faces from images. When getting an anime image, the faces from the images will be highlighted by drawing rectangles. Detecting whether the rectangle part of the image is an anime face by training the classifier(Support Vector Machine) with datasets(Histogram of Gradiants). And perform a sliding window algorithm & bounding box for a whole image.

3. Anime Face Recognition

When input an anime face image, we will get the information of the character from the face. We are planning to train and use the neural network (VGG16) to classify the anime faces by their labels from the datasets.