Unmanned aerial vehicle ship tracking method and system based on deep learning and difference value hash

Abstract

The invention provides an unmanned aerial vehicle ship tracking method and system based on deep learning and difference value hash, and the method comprises the steps: collecting unmanned aerial vehicle video data in a coastal region under visible light, extracting each frame of image; preprocessing based on the video image, and extracting a positive sample and a negative sample of the ship target; inputting a ship target sample in the video into a neural network through a region-based convolutional neural network mode, and carrying out model training; based on an improved difference value hash method, obtaining a ship candidate target result according to the model obtained by training, and calculating a hash value of each ship candidate target; and determining a final ship tracking resultat the current moment according to the hash values at the previous moment and the current moment. The method has a good detection result for complex scenes such as cloud, cloudy days and raining, ishigh in robustness, is better in stability, achieves the full automation of the tracking process, greatly improves the tracking speed and efficiency, and facilitates the application of an unmanned plane to the solving of a target tracking problem.

Description

technical field [0001] The invention belongs to the technical field of UAV ship tracking processing, in particular to a UAV ship tracking method and system based on deep learning and difference value hashing. Background technique [0002] In recent years, with the rapid development of computer networks, communications, and semiconductor technologies, UAVs, as a new type of remote sensor, have shown great potential in many fields such as environmental monitoring, disaster early warning, and animal protection. Due to its light and flexible features, it can fly at high speed and capture images or videos under many difficult conditions. UAV remote sensing technology, as an important means of spatial data collection, has the advantages of long life, real-time image transmission, high-risk detection, low cost, and strong flexibility, and is playing an increasingly important role in various fields. effect. [0003] Ship tracking technology is a focus of computer vision research, ...

AI Technical Summary

Problems solved by technology

The offline learning method is less effective for dynamically changing target tracking, while the online learning method is prone to error accumulation due to the introduction of new errors for each update, and eventually drifts or even loses the target.

Images