DeepLearning Based Action Recognition
Updated: Jul 6
Here at RidgeRun we believe in constantly researching and adopting new technologies into the solutions we provide. Therefore, we have been developing and incorporating the world of machine learning into our area of expertise. Particularly the field of activity recognition. This area of activity recognition in machine learning has found its way into more and larger industries worldwide by providing an automatic, reliable, and intelligent option to validate and complement production activities.
Inspired by the shift towards this new automated and intelligent industry, our R&D team has been working on an action recognition system that allows us to fulfill your needs with a smart and competitive computational system. However, the activity recognition task is not an easy task, since it is a spatio-temporal problem that differs from the commonly known classification or detection problems; hence, it requires to be tackled with newer and different techniques that incorporate time as a third dimension. Our system takes video samples as the spatio-temporal representation of an action, and based on previous training, it produces a multi-label output representing the likelihood of that action being each one of the training labels. This allows for detection, validation, and monitoring of sequentially dependent or independent tasks from any video source.
If you are looking to incorporate deep learning or any other automation technique into your project, do not hesitate to contact us, we will be happy to start working with you towards that goal you have.
Technical questions? : firstname.lastname@example.org
Want to know more about the project? : email@example.com