Unmanned aerial vehicle coastline floating garbage inspection system based on DANet

Abstract

The invention discloses a DANet-based unmanned aerial vehicle coastline floating garbage inspection system, which uses a panoramic segmentation algorithm to segment background and foreground target objects in an image at the same time and give an independent identity to each foreground target. The unmanned aerial vehicle is helped to adjust the course through an accurate segmentation result, a flight path is automatically planned, meanwhile, the floating garbage condition in the coastline is detected, pollution feedback positioning and types are found, and related departments are helped to solve the pollution inspection problem in a long coastline scene.

Description

technical field [0001] The invention relates to the technical field of inspection systems, in particular to a DANet-based inspection system for floating rubbish along the UAV coastline. Background technique [0002] As marine debris becomes more and more serious, pollution is likely to enter the human body with marine life, so reducing marine debris has become particularly important. Due to the size of the ocean, most of the garbage that is cleaned up at this stage is floating on the coast. However, the coastline is too long and the edges are more tortuous. It is difficult to draw an accurate coastline, and there is no accurate cleaning path. The coastline is long and tortuous, and it is even more unrealistic to clean up the entire coastline. Therefore, it is necessary to know where the garbage is and selectively clean up. In addition, some coastlines are difficult for people to reach, which affects the efficiency of manual detection of floating garbage. Traditional patro...

AI Technical Summary

Problems solved by technology

In addition, some coastlines are difficult for people to reach, which affects the efficiency of manual detection of floating garbage

Traditional patrol inspection methods include manual patrol, on-site monitoring

Manual patrols generally only inspect flat land, and it is difficult for people to inspect cliffs and rocky shores.

The inspection cycle is long, limited by the visual range, which consumes a lot of manpower and material resources; the cost of on-site monitoring layout is high, and it cannot cover the entire watershed, and the shooting range of monitoring equipment is limited, so it is easy to miss inspection

Surveillance video requires manual analysis, which further increases labor costs and financial costs. In addition, the effect of human analysis varies from person to person, and the analysis results are unstable. high

Although this method can automatically find the way for the UAV to fly, it still requires a lot of manpower to observe

Moreover, the dynamic binarization detection method has low robustness, and the actual coastline situation is changeable, which can easily affect the accuracy of the algorithm and make the UAV deviate from the ideal route.

Images