In Colombia, motorcyclists are the primary victims of traffic accidents. Between 2001 and 2014, there were about 28,000 deaths on the country's urban and rural roads; about half of these deaths occurred as a result of the lack of use of passive protection elements (safety helmets). This was accompanied by an increase in the number of motorcycles during those fourteen years of close to 445%. The large number of motorcycles and the absence of the transit authority from several municipalities have made it impossible to enforce compliance with traffic regulations for this population, and especially with the use of safety helmets. To solve this problem, an algorithm is proposed that uses image processing techniques in conjunction with artificial intelligence models for the detection of motorcycles and the use of helmets by the riders. The techniques used in this research for the detection of motorcycles within the vehicular flow include, among others, subtraction of the background, moment-preserving thresholding, morphological analysis and convolutional neural networks for the correct classification of the different objects found in the images. For the detection of helmets, a region of interest (ROI) is extracted from the original image and the contour is constructed within the posterior region of the ROI. Subsequently, the background of the image is subtracted and the H coordinate extracted from the HSB (hue, saturation and brightness) stack of the original RGB image. A classifier constructed using convolutional neural networks is used to determine the use of helmets by motorcyclists. The precision of motorcycle classification was 97.14%, whereas the precision of helmet detection was 85.29%. This algorithm, used with the equipment necessary for the identification of the vehicle's identity (i.e. LPR), can automatically assist in the enforcement task by public authorities in order to reduce the high mortality rate in this population group.