REAL TIME HUMAN DETECTION AND COUNTING

    INTRODUCTION

Computer vision is an interdisciplinary field that enables computers to understand the content of digital images at a high level. Computer Vision is the most used branch of AI (artificial intelligence) where machines and systems are able to get meaningful data from digital images or any other visual inputs. There are several types of computer vision such as image segmentation, object detection, facial recognition, edge detection, pattern matching, image classification and feature matching. Human detection is the task of finding all the occurrences of people in an image and has most commonly been achieved by scanning over every position in the image, at every scale, comparing a small window at each location with templates/patterns for what humans look like.

LITERATURE REVIEW

The development of real time human detection and computation systems is supported by a wealth of data from computer vision, artificial intelligence, and sensor technology. This review highlights key results, approaches, and advances in the field. Instantaneous tracking of individuals is essential to ensure accurate counting. Traditional techniques such as Kalman filters and mean transform algorithms, while effective, are limited by their inability to resolve and reidentify occlusions. However, continued innovation is needed to overcome current challenges and extend their applicability to more complex and sensitive domains. Future research should focus on ethical AI applications, improve computing performance, and develop large-scale solutions for a variety of real-world applications.

EXISTING SYSTEM

Human detection by instant detection and counting methods usually relies on traditional methods such as background subtraction, visual detection or simple machine learning. Although these methods have good results, it is difficult to guarantee their accuracy in complex environments such as occlusions, lighting changes and crowds. The scheme uses deep learning models such as CNN and techniques such as YOLO or SSD to improve the recognition of facts by learning the detailed features of human images. Unlike the old system, the scheme can handle dynamic and crowded environments while maintaining the processing speed, thus achieving a better balance between accuracy and efficiency.

PROPOSED SYSTEM

We propose an instant detection and counting system using YOLOv3 and OpenCV. The system will use cameras to capture videos, process them using YOLOv3 to detect people, and use OpenCV to count people. The system will provide a user-friendly interface for customizing and viewing real-time inspection data. We will conduct a feasibility study, create detailed design documents, and estimate the cost of the project. We will also ensure fairness and legality, including data privacy and security. The planning process can be used in many industries including retail, healthcare, transportation, event management and public sector organisations.

DESIGN

                            

                                          

RESULT

CONCLUSION

Real-time human detection and counting is a critical technology with wide-ranging applications in surveillance, crowd management, and retail analytics. While traditional methods struggle with accuracy and adaptability in dynamic environments, modern deep learning-based approaches, particularly using CNNs and object detection frameworks like YOLO and SSD, offer significant improvements in both detection precision and real-time performance.

REFERENCES

Al-Zaydi, Z., Vuksanovic, B., & Habeeb, I. (2018). Image processingbased ambient context-aware people. International Journal ofMachine Learning and Computing,8,268–273. https://doi.org/10.18178/ijmlc.2018.8.3.698

Al-Zaydi, Z. N., Ndzi, D, & Sanders, D. A. (2016). Cascade methodfor image processing based people detection and counting. InProceedings of 2016 international conference on imageprocessing, production and computer science (pp. 30–36).

Bernama. (2021, February 26). RM10,000 fine for SOP violationsbeginning March 11. Retrieved from newstraitstimes: https://www.nst.com.my/news/nation/2021/02/669100/rm10000-fine-sop-violations-beginning-march-11

Congrex. (2020, March 30). Disruption in the business eventsindustry: rising to the challenges of Covid-19. Retrievedfrom congrex https://congrex.com/blog/disruption-business-events-industry-challenges-covid-19