Bolt looseness detection algorithm and detection system based on improved twin neural network

The invention discloses a bolt looseness detection algorithm and detection system based on an improved twin neural network, proposes to use a twin network to perform bolt looseness detection, can perform looseness detection for bolts in different weather environments and at different acquisition angles, does not need to add additional marking lines, reduces the labor cost, and improves the detection efficiency. And the structure of the twin network is improved according to the task characteristics of bolt looseness detection, so that the looseness detection accuracy is improved. 本发明公布了基于改进型孪生神经网络的螺栓松动检测算法及检测系统，提出使用孪生网络来进行螺栓松动检测，可针对不同天气环境、不同采集角度的螺栓进行松动检测，且无须增添额外的标记线，降低了人工成本，并针对螺栓松动检测的任务特点对孪生网络的结构进行了改进，提高了松动检测的精准度。