Videos of children with Gowers' Sign and healthy ones

Gowers' Sign is a visual symptom exhibited by many neuromuscular dystrophies, including Becker muscular dystrophy, congenital muscular dystrophy, congenital myopathy, and Duchenne muscular dystrophy, which is the most aggressive, with a life expectancy of 20 to 30 years. Additionally, there is a 2.5-year gap between the onset of initial symptoms and a confirmed diagnosis. Early detection allows for the treatment of the disease, leading to a better quality of life. To the best of our knowledge, a non-invasive computer vision system for detecting Gowers' Sign has not yet been proposed. This research paper aims to demonstrate that convolutional networks with recurrent networks can effectively detect the Gowers' Sign maneuver in children aged 3 to 6 years. To achieve this, we created a video database with two classes: one with the Gowers' Sign maneuver and one without. Since the videos collected from various sources have different resolutions, we employed a method called padding to prevent distortions caused by resizing and rescaling during the convolutional network processing.