Convolutional Neural Networks for Automatic Identification of Individuals at Terrestrial Terminals

The study aimed to develop an automatic identification system based on convolutional neural networks (CNN) to detect wanted individuals in land terminals in Chiclayo, Peru, optimizing public safety through a faster and more accurate identification of persons with arrest warrants and strengthening the response capacity of law enforcement, to achieve this, used an applied technological approach with quasi-experimental design and method quantitative using tools such as Google Colab, Visual Studio Code and Python along with libraries like PyTorch and Keras to implement learning models deep while working with a private data set composed of 50,000 images processed using data augmentation techniques and evaluated the system performance with metrics such as precision, recall, F1-score and accuracy, the The results demonstrated an outstanding performance of the implemented models as InsightFace achieved an accuracy of 97.53%, a Recall of 100% and an F1 score of 98.76% while YOLOv8 obtained an accuracy of 95.93% and a Recall 99%, showing high effectiveness in recognizing and tracking people in real time, in conclusion the system met the objectives set for offer an efficient and reliable solution for automatic identification in land terminals, although areas for improvement related to the reduction of false positives in complex conditions which represents a step significant towards the modernization of surveillance systems in contexts with technological limitations.