This operation aims to improve the service provided by the Polícia Judiciária (PJ) to citizens by automating the operational process of cataloguing photographic images and comparing them with the existing image bank with tens of thousands of images, implementing a prototype that uses the latest Artificial Intelligence (AI) approaches. In particular, an innovative solution will be created for the ballistics area, identified by the PJ as a priority area, which will allow to identify with a high rate of success and speed the correspondence between projectile and possible type of weapon used, reducing the effort and intervention time of specialized human resources, which will have an impact in the process of evidence production and consequently will make Portugal a fairer country. The solution to be developed at ISCTE consists in the implementation of a specific classification model, based on Siamese neural networks, which allows the identification of weapon models from images of the casings of deflagrated projectiles (ballistic classification system). The training and testing of the model will make use of a previously created image database containing a large number of images of deflated projectile casings.