Anime-Recommender-with-your-Mood

Recommendation system plays an important role in various industries such as music applications, video platforms, food delivery and so on. After I learnt from different kinds of methods to implement recommendation systems, I consider that the limitation of these methods exist because the data and reference that they used in these methods do not include the user’s mood at the moment they are watching or using the product. Therefore, in this project plan, not only understanding some of the recommendation methods, I would also like to develop a more advanced recommendation engine by combining the real-time emotion detection and gesture recognition system with these methods.

Final report >> Click here

METHODOLOGY

Recommendation System Methods

1. Popular-based approach
2. Content-based filtering approach
3. Collaborative filtering using k-Nearest Neighbors approach

Emotion classification and detection

1. CNN model
2. vgg16
3. Resnet50
4. mobilenet

Result

emotion.detection.mp4

User gesture recognition

1. Gesture recognition

gesture_recognition.mp4