Crowd Management and Abnormal behavioure Detection

Artificial Intelligence and Machine Learning are going to be our biggest helper in the coming decade! Today morning, I was reading an article which reported that an AI system won against 20 lawyers and the lawyers were actually happy that AI can take care of repetitive part of their roles an

Project Title

Crowd Management and Abnormal behavioure Detection

Project Area of Specialization

Artificial Intelligence

Project Summary

Introduction

Artificial Intelligence and Machine Learning are going to be our biggest helper in the coming decade!

Today morning, I was reading an article which reported that an AI system won against 20 lawyers and the lawyers were actually happy that AI can take care of repetitive part of their roles and help them work on complex topics. These lawyers were happy that AI will enable them to have more fulfilling roles.

Today, I will be sharing a similar example – How to count a number of people in the crowd using Deep Learning And Computer vision? But, before we do that – let us develop a sense of how easy life is for a Crowd Counting Scientist.

Act like a Crowd Counting Scientist

Let’s start!

Can you help me count/estimate the number of people in this picture attending this event?  Click on the link to open the image.https://drive.google.com/open?id=1CTI4kPtmgVLrTWaRCOWGdkTxdwGcizeI

Crowd Estimation:

Crowd Counting is a technique to count or estimate the number of people in an image.

Crowd abnormal Behaviour Detection:

With the help of intelligent computer vision, we can differentiate what is systematically important in observing human interaction as normal behavior and abnormal behavior.

User Application:

We will design a native application for the user end and our app will provide these facilities to users/visitors.

For now, we are designing our system for Haram Shareef so we design our app to facilitate specifically for haram Shareef.  Which can be replaced according to project need where it is going to implement.

App Features are........

Finding the shortest path for the user according to the crowd situation.

Provide information about crowd status at a specific place. 

Provide the overall crowd flow. 

Predicted the crowd flow in the near future according to a specific event.

Update the user to the with different tips throw event Calendar.

Crowd Management:

With the help of these two modules, we are being about to manage the crowd, Also our system will help the management team to perform their task more efficiently and ensure any time of misadventure regarding stamped, overcrowded and also we will be able to boots client or their customer experience to visit this place through our user app as I mentioned earlier.

Project Objectives

Objectives 

• Extraction of point of interest from the frames of videos captured by surveillance cameras. 
•  
• In this case, the point of interest will be the individuals more specifically the heads of the individuals in the crowd.
• Detection of the head will further lead us to detect the normal and abnormal behavior in-crowd which will be done on the basis of speed and directions of individuals in-crowd.
• Calculating the number of head (Head counting) using any of these particular approaches some of the approaches are (Fourier approach, SIFT (Scale Invariant Feature Transform) approach and MRF (Markov Random Field).
• The benefits of headcount, on the other hand, are to find out the capacity of the crowd that can be fitted in the hall or in courtyard.
• As there are several entrance gates of Haram Sharif so when we meet a situation of having too many crowds at any one of the gates and the capacity of the hall is full in that case we will have a notification or a pop-up message that will let us inform about the situation. In order to handle the situation, we will close that particular gate and the crowd will be sent to another entrance gate.
• Crowd management will also avoid a greater number of accidents at Mina.
• Signal will be generated at the time of prayer insuring the management members to acknowledge the woman’s to leave Matta.
• The same goes for the ablution and washroom blocks. If anyone gets full then people will be informed to go to another nearest block.
• Through our app, a person can find the nearest gates, shortest paths to his destination from his location. But the app will be made if we got the results earlier for the main objective which is the detection of abnormal behavior.

Project Implementation Method

CNN-based methods

Ah, good old reliable convolutional neural networks (CNNs). Instead of looking at the patches of an image, we build an end-to-end regression method using CNNs. This takes the entire image as input and directly generates the crowd count. CNN's work really well with regression or classification tasks, and they have also proved their worth in generating density maps.

CSRNet, a technique we will implement in this article, deploys a deeper CNN for capturing high-level features and generating high-quality density maps without expanding the network complexity. Let’s understand what CSRNet is before jumping to the coding section

According to research  Paper

A regression guided detection network (RDNet) is proposed for RGB-D crowd counting and localization, which leverages a density map to boost the performance of detection for crowd counting. With the help of depth, i) a depth adaptive kernel is designed, which generates a high-?delity ground-truth density map and facilitates the regression-based crowd counting; ii) a depth-aware anchor is designed. Our depth-aware anchor facilitates the anchor initialization, and improves the detection of small heads; iii) even with point annotations, we can still use depth to estimate the sizes of bounding boxes, which shows their effectiveness for training RDNet. We further collect the large-scale ShanghaiTechRGBD crowd counting dataset for performance evaluation. Experiments on our dataset and MICC show that our method achieves the best performance for RGB-D crowd counting. Further, our method can be extended to RGB crowd counting and achieves comparable performance on the ShanghaiTech Part B dataset.

Benefits of the Project

 Benefits:

• Detection of abnormal behavior in the crowd
• Head counting.
• Crowd management

• There are plenty of other scenarios where crowd counting algorithms are changing the way industries work:

• Counting the number of people attending a sporting event
• Estimating how many people attended an inauguration or a march (political rallies, perhaps)
• Monitoring of high-traffic areas
• Helping with staffing allocation and resource allotment

Technical Details of Final Deliverable

The overall network architecture of our regression guided detection network (RDNet) for crowd counting is shown in Figure 2. It contains two modules: a density map regression module and a head detection module. In the density map regression module, a depth-adaptive kernel is introduced to generate a high-?delity ground-truth density map. In the detection module, we leverage a RetinaNet [14] for detection in view of its advantages in both speed and performance. We feed the estimated density map to the

classi?catension branch in the detection network to facilitate the classi?cation of heads, meanwhile, the depth-aware anchor strategy is also proposed to initialize appropriate anchor, which also helps the improvement of detection performance

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Security

Other Industries

IT

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Decent Work and Economic Growth, Sustainable Cities and Communities

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com