One of the causes of the high number of work accidents in Indonesia is not using personal protective equipment. The project helmet is a personal protective equipment that serves to protect he head. However, the level of awareness of workers using helmets in this project is still lacking. his study aims to determine the accuracy level of introduction to the use of standard project helmets. Scale Invariant Feature Transform (SIFT) and Support Vector Machine (SVM) are feature extraction and classification methods used in this study. The data used is in the form of 90 photos which are divided equally into 3 types of images. Research shows that there are 170 out of 180 upper bodies that have been successfully detected. The kernels used are linear, gaussian and polynomial. By sing 119 data as training data and 51 data as test data, the highest accuracy results are obtained the linear kernel with an overall accuracy rate of 68.63%.