基于Faster R-CNN的标识识别系统

Sign Recognition System Based on Faster R-CNN

随着工业机器人技术的不断推进,在制造业中依靠机器人参与的自动化生产线得到广泛应用。为适应市场个性化需求,对于柔性程度高、相似件经常更换的生产线,需要机器人自主识别待加工零件的种类,减少对人工的依赖,避免人为的错误。针对生产线上工件的自动识别问题,本文提出一种基于深度卷积神经网Faster R-CNN的视觉检测方法并应用到3种汽车前桥标识识别中。试验结果表明,本文方法可以用于实际生产过程中。

With the continuous advancement of industrial robot technology,automated production lines that rely on the participation of robots in the manufacturing industry are widely used.In order to meet the personalized needs of the market,for the production line with high flexibility and frequent replacement of similar parts,it is necessary for the robot to independently identify the types of processing parts,so as to reduce the dependence on labor,and avoid human errors.Aiming at the automatic identification of workpieces on the production line,this paper proposes a visual detection method based on convolutional neural network Faster R-CNN in deep learning and applies it to the identification of three kinds of automobile front axle marks.The test results show that the method proposed in the paper can be used in the actual production process.