Cause analysis of incorrect time of AI traffic statistics of Dahua Equipment Development Bank

Traffic statistics are based on AI Artificial intelligence algorithm , It can count the number and flow trend of human bodies in video images in real time , Combined with video surveillance technology , It can realize the statistics of people flow in the current monitoring area , Understand the dynamics of pedestrian flow in real time , It can also achieve crowd density monitoring 、 The purpose of avoiding crowd gathering , It plays an important role in security and other scenes .TSINGSEE Qingxi video has also been actively developing AI Integration of intelligent detection technology and video capability , Including face detection 、 Face recognition 、 Vehicle detection 、 License plate recognition and so on .

Share today's tips in development ： The problem of incorrect time to obtain pedestrian flow statistics through Dahua equipment is solved .

When obtaining pedestrian flow statistics , The information returned at the statistical time point is wrong . year / month / Japan / when / branch / Second information is incorrect , Such as ： year （dwYear） Is shown as 15, Here's the picture ：

1） First, compare the official demo, Whether there are exceptions in the code .

official demo example ：

Encapsulated into library code ：

As can be seen from the above code , There is not much difference between the two codes . So why is it that the query time point of statistics is incorrect ？ Both code break point debugging , Only the of the package library can be used to count pedestrian flow , A bit of a timing error has occurred .

Since there's nothing wrong with the code , Then it is possible to call the wrong version of the library . Check whether the number of bytes in the library is the same , Then check the version of the library .

The official library dll：

The encapsulation library uses Dahua dll：

By doing dll By comparing and analyzing the number of bytes, it is concluded that ：avnetsdk.dll and dhconfigsdk.dll Wait, there's a mistake , With the official demo Of dll Mismatch .

terms of settlement ： Will official demo All of the dll Copy them all , Put it in your own program and run it again , At this time, the time obtained by the traffic statistics has been completely correct .

With the continuous development and application of artificial intelligence technology , Security video platform and AI The combination of technology has become the trend of market development .TSINGSEE Green rhinoceros video has also recently released edge based AI Computing hardware device ——AI Safety production camera , The device adopts a new embedded multi algorithm framework software , Built in a variety of AI Algorithm , Enterprises can choose the algorithm according to the camera configuration , At present, it can support helmet detection 、 Pyrotechnic detection 、 Indoor channel blockage detection 、 Off duty sleep detection 、 Personnel intrusion detection 、 Perimeter intrusion detection 、 Occupancy detection of outdoor fire passage, etc , It is of great significance in the supervision of enterprise safety production . Users interested in our video platform can go to the demonstration platform for experience or deployment testing .