Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 2

FLOWCHART:

 The Input information is parted into preparing and testing sets.


 A deep learning-based recognition worldview is utilized to recognize people in groupings.
 Because of the best exhibition results for conventional object recognition in this work, YOLOv3 is
utilized.
 The model used single-stage network design to estimate the bounding boxes.
 The model was initially prepared on the COCO informational index. For overhead view individual
recognition, upgrade the detection model's productivity, and another layer of overhead
preparing is added with the current design.
 After identification, the bounding box data for the most part centroid in arrangement, is utilized
to register each bounding box centroid distance.
 We utilized Euclidean distance and determined the distance between each identified bounding
box of people groups.
 Following figuring centroid distance, a predefined edge is utilized to check either the distance
among any two bounding box centroids is not exactly the arranged number of pixels or not.
 If the two individuals are near one another and the distance value violates the base social
distance edge then the information stored in violation set and the color changed to red.
 Centriod tracking algorithm is used to track detected humans and show the total information of
total number of violations.

BOUNDING BOX:

 It is a rectangle that serves as a point of reference for object detection.


 And creates a box for that object.
 It is used to detect object or target locations.
 Data annotators draw these rectangles over images, outlining the object of interest within each
image.
 Determined by X and Y coordinates. one for upper right corner and other for lower right corner.
 The ultimate goal for object detection task is to enable the model predict the bounding box as
closed as possible.

CONCLUSION:

This research presented a surveillance system for people tracking and social distancing classification
based on thermal images. The pre-trained YOLOv3 model is used for human detection. A specific
algorithm was implemented on bounding boxes to distinguish between safe and unsafe conditions,
respectively, marking as green and red the bounding box for detected persons. The proposed approach
can be implemented in a distributed video system and also in Images. In the future, we will utilize this
methodology on mobiles and other Gadgets. This technique achieved promising results for people
detection in terms of evaluation the accuracy and precision of the detector comparable to the other
deep learning models.

You might also like