Deep Learning Based Object Detection in Surveillance Video

Object Tracking from surveillance video has become a popular research area in the fields of image processing, video processing, and computer vision. The abilityto accurately identify specific objects from surveillance footage can have a significant impact on public safety and security. Our research employs YOLOv5, a state-of-the-art deep learning algorithm, for object detection in this domain. YOLOv5 is particularly effective in identifying objects in real time with high accuracy and speed. In this paper, we focus on using YOLOv5 for the detection of common objects in our daily lives. With various crimes and thefts occurring in society, it is challenging to guarantee everyone's safety on the streets, highways, or roadsides. Therefore, our research aims to address this issue by exploring how we can reduce manual labor and improve safety through the detection of specific objects in different scenarios. For instance, we can identify individuals based on their attire or if they are carrying a weapon or dangerous item. By fine-tuning our model, we can accurately detect and identify the required objects, allowing us to take appropriate action and enhance public safety. Overall, our research shows the potential of YOLOv5 for object detection in specific scenarios and demonstrates how it can be used to enhance public safety and security.