Identification of Skin Color Images in Humans Using Principal Component Analysis Method

Abstract

Skin detection is an important process for classifying human skin color. Some phone cameras produce images in RGB (Red, Green, Blue) format. Various cases of skin detection often perform a transformation from the RGB color space to another color space, such as YCbCr or HSV, to improve detection accuracy. In this journal, we conduct research on skin detection using Teachable Machine, a platform that utilizes machine learning to teach computers to recognize visual patterns. The term "accuracy per epoch" refers to how well the model is at telling which pixels in an image are skin color during each training iteration (epoch) when skin color identification using the Principal Component Analysis (PCA) method on a teachable machine is being done. PCA helps reduce the dimensionality of data, enabling more efficient analysis and increased model accuracy. Teachable Machine makes this process simple by providing a user-friendly interface for training models without requiring in-depth knowledge of code or algorithms. Measuring accuracy per epoch is crucial in the training process because it shows the development of the model's ability to detect human skin over time. We expect the model to improve its accuracy in identifying skin-color pixels with each epoch, thereby enhancing the overall performance of the skin detection system. This research shows the great potential of the teachable machine in skin detection applications, providing a more efficient and accurate solution than traditional methods.