Learning something without a mentor is difficult. Many a times financial, cultural and geographical barriers restrict learners to learn from good mentors. AI can be used to help these learners and dance lovers to improve at dancing.
This AI model is trained on dance movements of expert dancers and then used to predict dance moves by analyzing past and current dance moves of learner.
We have used Google Cloud Video Intelligence API to analyze movements of dancers by calculating relative positions of their body parts thus describing the posture using various heuristics. Using geometry and simple metrics we have tried to compare the similarity of body movements and thus recommend moves on basis of similarity.
We have researched a lot on analyzing body movements and describing body postures using ML and AI. We had to design metrics to find how similar two body postures are. We have used Deep learning to make videos for testing this AI model.
We have build a very sophisticated AI model. The result of these models are really impressive. It detects similarity between two moves up to a great extent and the metrics have been working fine.
We have learned concepts of advanced AI, geometry and practical experience of using rest api.
We are trying to use music and dance together to improve dance moves recommendation. We are even working on improving metrics that determine similarity of two movements.
Built With
- ai
- geometry
- google-cloud
- ml
Log in or sign up for Devpost to join the conversation.