Research on object detection of robot based on convolutional neural network

At present, robot target detection is widely used in tremendous fields including industrial, home, and commercial robots. However, the existing detection methods have some problems, such as low recognition accuracy, slow processing speed, and poor adaptability to the environment. In this work, an object detection model based on deep learning was designed, which mainly included a feature extraction network, a regional suggestion network, and an object classification and localization network. The feature extraction network uses the pre-trained ResNet-50 as the backbone network to extract the high-dimensional features of the image. The region proposal network (RPN) is responsible for generating high-quality candidate regions, which are then fed into subsequent classification and location networks. The target classification and localization network utilize multilayer perceptron (MLP) to perform accurate target category classification and bounding box regression for candidate regions. Experimental results show that the proposed model is significantly better than the traditional target detection methods in terms of accuracy and real-time performance.