Search

Automatic population density estimation is a significant study area in intelligent video monitoring. Traditional methods need design features manually, which are hard to keep pace with the current state of big data. At the same time, with the outbreak of artificial intelligence methods such as deep learning, the application of deep learning to video monitoring is also an irresistible trend. Therefore, according to the disadvantage of traditional manual feature extraction and the deficiency of single-layer convolutional neural network(CNN), a multilayer convolutional neural network(MCNN) is raised. In this article, head size changes caused by various reasons, such as penetration effect, will not affect the characteristics of CNN learning pictures. That is to say, even if we do not know the perspective of the input map, we can accurately detect the population density on the basis of adaptive kernel. The characteristic graphs of each layer are integrated to obtain the population density map. experiments reveals that this network structure can attain more accurate population estimation.

An effective smoke detection from visual scenes is crucial to avoid large scale fire around the world. But it is still challenging due to its large variations in color, texture and shapes. To improve smoke detection accuracy, a new approach based on deep convolutional neural networks is proposed which can be trained end to end from raw pixel values to classifier outputs and automatically extract features from images. Experiments show that this method achieves 99.4% detection rates with 0.44% false alarm rates on the large dataset which obviously outperforms existing traditional methods.

A review of deep learning based soft computing techniques in several applications is presented. On one hand, soft computing, defined as a group of methodologies, is an important element for constructing a new generation of computational intelligent system and has gained great success in solving practical computing problems. On the other hand, deep learning has become one of the most promising techniques in artificial intelligence in the past decade. Since soft computing is an evolving collection of methodologies, by presenting the latest research results of soft computing based on deep learning, this review not only reveals a promising direction for soft computing by incorporating deep learning, but also gives some suggestions for improving the performance of deep learning with soft computing techniques.

To evaluate and analyze the performance of the rule "keep right except to pass", a model of traffic flow on a multi-lane freeway based on cellular automaton (CA) is proposed. Taking lane-changing frequency as the main indicator of traffic safety level, we analyze the impact of different lane changing and deceleration probability on the traffic flow and safety in light and heavy traffic. And then we compare the "keep-right-except-to-pass" rule with the free rule in the two-lane case. The results demonstrate that "keep-right-except-to-pass" rule is not as effective as the free rule in promoting traffic flow, however, this rule ensures better safety for drivers than the free rule. Additionally, a new traffic rule, which sets different posted speed limits for adjacent lanes, is designed to promote a better traffic flow with the safety requirements satisfied.