BlogCases

Our team helped to launch dancing app MVP using computer vision algorithms

With this dancing app users can learn how to dance with interactive tutoring.

Business Task

The client needed development of the core of the application to detect person's movement and compare them with correct movement provided in the tutor and give it a score to provide feedback. Client also required consulting on building the architecture of the product and guidance on the possibilities of ML/AI to implement the whole solution.

Solutions

Our team used high-accuracy pose-estimation algorithm to detect poses of the person. We've built a sequence matching algorithm to provide relative and accurate feedback to the users. We've gone all the path from tech specification to deployment of the application to AWS.

Value delivered by Dysnix

Our team participated in building the strategy of development of ML/AI part of the product. We've implemented ML/AI parts of the product. We've deployed and support infrastructure on the AWS.

Conclusion

We've launched an MVP successfully and it's being tested at the moment to find product-market fit. We continue working on the project until its release.

Industry
Location
Project duration
Our team
Technologies used

PyTorch

Human pose estimation

Detectron

Faiss

Numpy

Docker

Services provided

AWS