A ship smoke emission online detection and tracking method based on deep learning

Abstract

The invention discloses a ship smoke emission online detection and tracking method based on deep learning, and the method mainly comprises two parts of smog detection and smog tracking, firstly, converting ship smoke video data collected in Wuhan river basins and Shenzhen harbors in Yangtze River into standard data set formats for training a detection model and a tracking model; secondly, respectively constructing a detection network and a tracking network by using TensorFlow, setting hyper-parameters such as the learning rate, the number of batch samples, the number of training iterations andthe like, then loading a training data set to start training, and obtaining a detection and tracking model after the training is completed; and finally, inputting a video clip or an online video stream into the detection model, stopping the detection process and setting the coordinate information and the frame which detects the smoke for the first time as initial position information and a template image which are input into the tracking model for tracking when detecting that the smoke exists in a certain coordinate area in a certain frame. Compared with the prior art, the method of the invention has the beneficial effects of low cost and high time efficiency.

Description

technical field [0001] The invention belongs to the technical field of intelligent green transportation, and relates to an online detection and tracking method for ship smoke emission, in particular to a deep learning-based visual detection and tracking method for ship smoke. technical background [0002] With the rapid economic development and the continuous growth of trade volume, my country's waterway cargo transportation volume and port throughput have ranked first in the world for many consecutive years. Busy water transportation and shipping bring economic benefits, but also air pollution, but waters are generally far away from the city center, and the air pollution it brings is easily overlooked. In fact, in the Beijing-Tianjin-Hebei, Yangtze River Delta, Pearl River Delta and coastal areas, ship smog emissions have become one of the important sources of air pollution. The fuel used by ships is mainly residual oil or heavy oil, which are all diesel, and the sulfur co...

AI Technical Summary

Problems solved by technology

Because the feature types of smoke are relatively complex, if the method of artificially constructing features is adopted, the time cost spent in the feature engineering stage is relatively high. In addition, because the artificially constructed features are limited by the limitations of manual experience, it is difficult to extract Reflect the more essential deep feature information of smoke, and then affect the effect of smoke detection

At present, there are many application cases of using this method for fire smoke detection, mainly forest fires, and the application of ship smoke emission detection is still in its infancy

Images