Smoke target detection based on dynamic multi-frame decimation with cavity convolution

Abstract Compared with traditional smoke alarms, detection based on video images has the advantages of low cost, fast response time, wide applicability, etc. Smoke algorithms are becoming an effective and widely used tool for early warning fires by recognizing graphic features. In this paper, we use VGG-16 neural network as benchmark improve it using null convolution combining dynamic multi-frame subtraction method detection. The experimental results show that proposed is useful in stage a fire before becomes intense.