YOLOv3 Algorithm to Measure Free Fall Time and Gravity Acceleration

Abstract

Computer vision methods as an alternative to sensors in modern measurements are feasible in physics experiments due to their speed, accuracy, and low cost. The You Only Look Once (YOLO) algorithm is widely used in computer vision because it detects object positions quickly and accurately. This research uses YOLO version 3 (YOLOv3) to compute an object’s falling time and gravitational acceleration. Two steps are performed in this study: first, the detection of predefined objects using YOLOv3, and second, the use of trained YOLOv3 to track the object's coordinate. According to the object tracking results, the object's falling time can be measured based on the object tracking results. The gravitational acceleration is calculated using the time data after the fall time of the object is measured. The measurement result of the fall time of the object will be compared with the data from the sensor. The result of the gravitational acceleration calculation is measured for its relative error against the value of 9.78150 m/s², which is the value of gravitational acceleration in Jakarta city. The results show that YOLOv3 can accurately detect objects and measure free-fall motion, with a time measurement error of only 1.1 milliseconds compared to sensor measurements. The error obtained from the measurement of the Earth's gravity is 0.634%.