US 12,356,064 B2
Measurement assistance system and method
Chun-Chih Kuo, Kaohsiung (TW); Chia-Hung Chang, Kaohsiung (TW); Bo-Yun Hou, Kaohsiung (TW); and Cheng-Yu Yang, Kaohsiung (TW)
Assigned to NATIONAL KAOHSIUNG UNIVERSITY OF SCIENCE AND TECHNOLOGY, Kaohsiung (TW)
Filed by National Kaohsiung University of Science and Technology, Kaohsiung (TW)
Filed on Dec. 2, 2022, as Appl. No. 18/073,617.
Prior Publication US 2024/0187722 A1, Jun. 6, 2024
Int. Cl. H04N 23/61 (2023.01); G03B 17/56 (2021.01); G06F 18/213 (2023.01); H04N 23/90 (2023.01)
CPC H04N 23/61 (2023.01) [G03B 17/56 (2013.01); G06F 18/213 (2023.01); H04N 23/90 (2023.01)] 18 Claims
OG exemplary drawing
 
1. A measurement assistance system, comprising:
a measurement platform, having an operation area configured for a to-be-measured object and at least one measurement tool to be placed;
at least one camera, arranged on the measurement platform and configured to obtain a measurement image; and
a server module, electrically connected to the at least one camera and configured to: execute a measurement tool identification program, a measurement part identification program, and a measurement posture identification program according to the measurement image through a standard measurement tool appearance model, a standard measurement part model, and a standard measurement posture model, obtain a measurement tool appearance image corresponding to the at least one measurement tool, a measurement part image of the to-be-measured object, and a measurement posture image of a measurer, and determine whether the measurement tool appearance image, the measurement part image, and the measurement posture image are correct, wherein
the server module has a processing unit, and when the measurement tool appearance image, the measurement part image, and the measurement posture image are all correct, a measurement result is generated according to measurement data; and
the standard measurement tool appearance model, the standard measurement part model, and the standard measurement posture model are trained through a pre-built deep learning neural network framework, wherein the deep learning neural network framework comprises a TensorFlow object detection algorithm, a Hu Moment algorithm, a TensorFlow CNN algorithm, and a MediaPipe Hand algorithm.