Crowd Analysis

Project Title

Crowd Analysis

Project Area of Specialization Computer ScienceProject Summary

In the rapidly moving world, in which we are living, the need and requirement of getting better and better is constantly increasing, due to which people observe doing crowd counting an issue in terms of time as they think that manually sorting would take a lot of time, energy and resources in terms of labor. Many solutions came in past which helped in doing analysis but every solution had some deficiencies or incompleteness for example some of the systems were not having enough precision and accuracy, some of them was time taking, some of them was not using proper techniques. Some of the techniques that are being used like detection-based methods, regression-based methods and density estimation-based methods have some deficiencies for example in detection based method only works if the image is not crowded, in regression based methods only low level features are extracted and in density estimation we first have to crop patches from the image and then for each patch low level features are extracted. However, our proposed solution provides an overall solution to above mentioned problems and issues. It includes using of proper technique i.e. Convolutional neural networks. In this method instead of looking at the patches of an image, we will build an end-end regression method using Convolutional neural networks. The entire image is taken as input and the count is then generated. These all features help us to provide a better and much accurate solution for counting.

Project Objectives

The world is moving on a very fast pace towards advancement in technology, the need to save time is a big problem for many and for that reasons different algorithms have developed and are developing. The algorithm is basically judged by its precision, accuracy and output time. To improve these factors new algorithms are constantly developing.

Our aim is to use such algorithm which can give us the precision as well as save our time. It can be used from small businesses to big industries. Data scientists and such algorithms are currently being used to do Crowd Analysis which is costly, time taking and the accuracy is not up to the mark. In our project our aim is to make a program which can be used by anyone as well as the cost is low and the estimation of the crowd will be

80-90 percent which is huge rises in percentage as compared to manual counting and the algorithms which were previously used. To help big industries our project will be able to calculate sets record means that multiple images can be given to the program and it will generate results combining all the images. Our project will also be capable of generating results from the video. The project will have a user-friendly interface which will help people understanding how to use the application without going into help section.

Project Implementation Method

We will be using different datasets to train our model. Once the model is trained using large-scale training sets, they can be applied to different crowd scenes without being trained again. Crowd Analyzer will be able to count number of people when an image or video will be given. Some Deep learning and Machine learning techniques will be used to perform these functions. The proposed application will be able to calculate results from images, video and live video. The application will have an upload bar where an image or video can be uploaded and the result after calculation will be uploaded in result bar. Besides this, the application will have the login/sign up page. Administrator will have access to the user database while the user will only be able to use the application without having administrator privileges.

Benefits of the Project

Crowd Analyzer will be helpful in many different ways for example in a sporting event the number of people can be counted as well as it can be used for monitoring of high traffic areas.

Technical Details of Final Deliverable

For software we will be using android studio and Anaconda software. We will be using Python and Java language. Paid software will not be used.

Final Deliverable of the Project Software SystemCore Industry ITOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	65000
Laptop	Equipment	1	40000	40000
External HDD	Equipment	1	15000	15000
Logitech webcam	Equipment	1	10000	10000