A mobilized automatic human body measure system using neural network

Mobilized automatic human body measurement systems possess high mobility, easy operation, and reasonable accuracy. However, existing systems focus on accuracy and robustness rather than mobility and convenience. To overcome this shortcoming, this work presents a mobilized automatic human body measure system using a neural network (MaHuMS-NN) to promote general measurement results by supervised learning. MaHuMS-NN based on general regression NN (GRNN) selects an image, performs image processing, segments the image, and detects a silhouette for feature point extraction in the silhouette. The system measures feature size. The significant contributions of this work are as follows. First, MaHuMS-NN is the first intelligent system for anthropometry in the Android platform. Second, unlike existing systems, MaHuMS-NN can intelligently adjust when the model is optimized for prediction and perform self-error correction based on individual characteristics. Experimental results indicate that compared with existing systems, MaHuMS-NN demonstrates better performance with an accuracy of less than 0.03 m.