Black smoke vehicle detection method based on convolutional attention network

Abstract

The invention discloses a black smoke vehicle detection method based on a convolutional attention network. The method comprises the following steps: extracting a moving target through a monitoring video; extracting a foreground target of the moving target, and removing a non-vehicle target in the foreground target; extracting a candidate black smoke area of the moving vehicle, and converting the candidate black smoke area into a set specification; extracting static features of the single-frame picture of the candidate black smoke area by using a convolutional attention network; inputting the static characteristics into a full connection layer to identify a black smoke vehicle; when the static features of the candidate black smoke area are extracted, the feature map carries out weight distribution through the channel attention module and the space attention module. According to the method, an attention mechanism is adopted to enhance the representation capability of the convolutional neural network in two dimensions of space and channel, pay attention to the main characteristics of the black smoke area and inhibit unnecessary characteristics of interference areas such as vehicles, roads and shadows, and the misjudgment rate can be effectively reduced.

Description

technical field [0001] The invention relates to a smoky car detection technology, in particular to a smoky car detection method based on a convolutional attention network. Background technique [0002] Smoky vehicles emit a large amount of fine particulate matter (PM2.5) and toxic gases (CO, NO, etc.) during driving, causing environmental pollution and seriously endangering human health. [0003] The traditional detection methods of smoky vehicles have problems such as high cost and low efficiency. With the rapid development of the Internet of Things and artificial intelligence (especially computer vision), video / image recognition algorithms are increasingly mature, and it is possible to automatically identify smoky cars based on surveillance videos. There are two types of smoky car detection methods based on computer vision, one is based on traditional machine learning algorithms (support vector machines, neural networks, etc.), and the other is based on convolutional neur...

AI Technical Summary

Problems solved by technology

However, in the existing technology, in the extracted target vehicle picture, in addition to the black smoke area, it also includes interference areas such as vehicles, roads, shadows, etc., and the feature information of the interference area extracted by the convolutional network may lead to misjudgment

Images