Excavator detection method for unmanned aerial vehicle routing inspection based on CenterNet

• Main Authors: YANG QINMIN, FAN HAIDONG, ZHOU JUNLIANG, CHEN JIMING, YU JIN, TENG WEIMING, QIAN WEIBIN, LI QINGYI, DING NAN, WU YUN, QIAN JIREN
• Format: Patent
• Language: Chinese; English
• Published: 02.10.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention relates to an excavator detection method for unmanned aerial vehicle inspection based on CenterNet. The excavator detection method comprises the following steps: S1, establishing a training data set; S2, carrying out training of the Center Net; S3, carrying out excavator detection. According to the method, a spatial attention mechanism is designed for a vehicle body and a bucket of the excavator, and corresponding marks are newly added on the format of a traditional target detection data set; a spatial attention mechanism is trained in combination with the annotation thermodynamic diagram to enable the spatial attention mechanism to be focused on the vehicle body and the bucket respectively, thereby realizing modeling of the importance of the spatial position in the input image; the visual characteristics of the excavator are fully utilized, the network is focused on the important area of the target, the influence of irrelevant backgrounds on the detection result is reduced, and then the detecti