Application of genetic algorithm for broad learning system optimization

In order to solve the problem of waste of resources and low monitoring efficiency caused by Container Freight Station (CFS) safety helmet detection mainly relying on manual operations. The Broad Learning System (BLS) optimized by Genetic Algorithm (GA) is selected as the image recognition classifier for helmet detection, which accurately and quickly marks the images of CFS workers who do not wear safety helmet in the video and issues an early warning. GA is a method of searching for the optimal solution by simulating the natural evolution process. While comparing it with initial BLS, BLS optimized by GA (GA-BLS) can automatically constructs network structure and tunes hyperparameters. Furthermore, using proposed method in safety helmet data set reaches a lower error, reduces operation time than initial BLS and other image detection method, such as support vector machine (SVM).