In this study, we propose a pedestrian attribute recognition method based on background suppression to solve the problems of background clutter and object occlusion associated with the monitor scene. The proposed method can reduce the impact of the background on pedestrian attribute recognition. First, three branches are generated by improving the convolutional neural network. These three branches are used to extract the features of the pedestrian images, human body regions, and background regions. Then, the regional contrast loss function and weighted cross-entropy loss function are considered to constitute the joint cost function of the network. The features learned by the neural network exhibit background clutter invariance under the constraint of the joint cost function. Therefore, the proposed method can improve the pedestrian attribute recognition accuracy. The proposed method was verified using the PETA and RAP pedestrian attribute datasets. The results denote that the proposed method exhibits improved the mean accuracy, accuracy, precision, and other performance indicators when compared with those exhibited by the remaining methods, confirming its effectiveness.