Method for generating target detection football candidate point of Nao robot based on Heatmap

Abstract

The invention relates to a method for generating a target detection football candidate point of a Nao robot based on a Heatmap. The method comprises the steps of selecting a convolutional neural network as a target detection model; simulating competition environments, collecting a plurality of groups of pictures to make a data set for training and testing, generating a Heatmap, processing to obtain a Heatmap visualization result, reconstructing the convolutional neural network to accelerate the network calculation speed, setting a proper threshold value, taking points greater than the set threshold value in the Heatmap as candidate points of a ball, and finally sending the candidate points into a classifier to obtain a final accurate identification result. According to the method, the adaptive capacity of the Nao robot vision system to the light environment of the competition field is enhanced; high-precision recognition of the football can be achieved in different light environments;feature extraction is completed through few convolution layers; the recognition real-time performance is guaranteed; and meanwhile the football recognition accuracy is greatly improved through the method that football candidate points are generated and then enter the classifier to be recognized.

Description

technical field [0001] The invention relates to a method for generating football candidate points for Nao robot target detection based on Heatmap. Background technique [0002] The Nao robot is a humanoid robot with powerful functions and high research value. It is used in education and research institutions all over the world. The Platform Group is a group dedicated to Nao robots playing humanoid soccer matches. In the game, Nao robot's recognition of football involves the decision-making of the game and the final outcome. The importance is self-evident, and it is a key research topic in its vision direction. Limited by the computing power of the robot's hardware equipment, the current solutions are: [0003] 1. Use the traditional method to generate ball candidate points and then classify to obtain the ball position. When generating ball candidate points, there will be large non-green gaps between the green pixels that meet the vertical scanning line and the non-green ga...

AI Technical Summary

Problems solved by technology

[0006] 1. The traditional method has the characteristics of small amount of calculation, high real-time performance, and simple implementation principle, but its robustness is low, and it is greatly affected by changes in the external lighting environment. It is easy to cause misjudgment and produce many wrong candidate points, thereby increasing the Afterwards, the calculation amount of classifier recognition makes the accuracy rate of the whole football recognition process low

[0007] 2. The method of directly using deep learning target detection is limited by real-time performance. In order to pursue real-time performance, the precision is sacrificed, which cannot show very good accuracy, and there are also many misidentifications.

Images