Text
IMPLEMENTASI METODE YOLO DAN DEPTH-SENSING TERHADAP MOTION TRACKING 3D PENDETEKSIAN JAGA JARAK UNTUK MENGURANGI RESIKO PENULARAN VIRUS COVID-19
The Activity of avoiding physical interaction between two people or more within 1.5 meters, known as Social Distancing, is one of many methods to prevent the spread of infectious disease. An infection rate of illness in the COVID-19 pandemic period is uncontrollable, specifically when the virus is new. Social distancing during the pandemic period has proven to minimize the infection rate of the COVID-19 virus. This research aims to create an application to detect the distance between all the people the camera sees using the YOLOv4 and the Depth-Sensing method. A challenge that needs to be faced is a condition when the computer thinks that the distance between two adjacent bounding boxes is close. But they are far apart because the first object is closer to the camera, while the second object is far behind the first object. The Depth-Sensing method detects the distance in the 3D domain using the depth similarity equations. Training and Testing of YOLOv4-tiny using CrowdHuman data shows a result of Mean Average Precision or mAP around 70.11% on CrowdHuman data and 99.40% on captured data with a training time of ±7 hours. Meanwhile, the average frame rate score is around 23 FPS for the Apple M1 CPU and 215 FPS for Nvidia Tesla T4 GPU. Implementing a social distancing detector using the depth-sensing method gave an average accuracy of 86.39%, a precision of 81.65%, and a recall of 85.77%.
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2207003849 | T79436 | T794362022 | Central Library (Referens) | Available but not for loan - Not for Loan |
No other version available