|
ArUcoE: Enhanced ArUco Marker |
|
Oguz Kedilioglu(FAU Erlangen-Nürnberg, Germany), Tomás Marcelo Bocco(Politecnico di Torino, Italy), Martin Landesberger(Technical University of Munich, Germany), Alessandro Rizzo(Politecnico di Torino, Italy), Jörg Franke(FAU Erlangen-Nürnberg, Germany) |
| This paper presents a novel fiducial marker type called ArUcoE. It is obtained from a standard ArUco marker
by enhancing it with a chessboard-like pattern. With our approach the pose estimation accuracy of any ArUco marker can easily be increased. Further methods to increase the accuracy are analyzed. By applying a subpixel algorithm to the corner regions we are able to locate the corner points within a pixel and overcome the restriction of pixel-level accuracy. A deeplearning-based super-resolution method is used to artificially increase the pixel density in the same regions. |
|
|
Self Evaluation for Gait based on Optical Flow Calculation |
|
RYO AKAGI, Teruo Yamaguchi(Kumamoto University, Japan) |
| Walking has the effect of preventing diseases and unhealthy conditions that are increasing in many people. At present, it is common for experienced and knowledgeable instructors to evaluate gait in walking, but it is not certain that the evaluation is always objective, and it is time-consuming and costly. Therefore, we expected that if we can analyze the ideal gait and compare our own gait with the ideal gait. we would be able to evaluate our own gait in a simple way. In this study, to analyze the ideal gait, we measure the head velocity and acceleration of the ideal gait captured by a high-speed camera and compared it with the non-ideal gait using optical flow measurement. |
|
|
Visualization of Torque Generated in the Human Body based on Optical Flow Measurement |
|
Yuta Yamaryo, Teruo Yamaguchi(Kumamoto University, Japan) |
| Athletes are constantly looking for body motion that lead to better results. However, in general, the torque and force generated by the body motion are invisible, so direct measurement is difficult. Therefore, we thought that the motion can be used as an input to If we can achieve this, it will be possible to compare difference with experienced and beginners in movement, which will help the athletes improve athletic skills. In this paper, we examined the possibility of visualizing changes in torque that occur during rotational movement with one degree of freedom. Experimental results showed that the possibility of visualizing changes in torque during exercise can be achieved. |
|
|
Performance Evaluation of YOLOv3 and YOLOv4 Detectors on Elevator ButtonDataset for Mobile Robot |
|
Sumaira Manzoor(Sungkyunkwan University, Korea) |
| The performance evaluation of an AI network model is the important part for building an effective solution
before its deployment in real-world on the robot. In our study, we have implemented YOLOv3-tiny & YOLOv4-tiny darknet based frameworks for performance evaluation of the elevator button recognition task. Our two-fold object is: to overcome the limitation of elevator buttons dataset by creating new dataset; to conduct the performance evaluation of both detectors using the precision, recall, F1-score and mAP metrics. The purpose of our work is to assist the researchers and developers in decision making of suitable detector selection for deployment in elevator button recognition application |
|