M Linna, J Kannala, E Rahtu - Proceedings of the 13th International …, 2018 - oulurepo.oulu.fi
… than many of the state-of-the-art methods. Our method can be thought of as a replacement for Kinect … most likely these false positives could be filtered, especially with use case specific …
… Each convolutional layer have different filter size and they … for Kinect [23] by using a fast and accurate pose estimation … Humanposeestimationusingbody parts dependent joint …
… layers with filters of 3×3 and two layers with filters of 1×1, … a Kinect2 and exhibiting one or multiplepeoplepeople passing in … Realtime multi-person 2d poseestimationusing part affinity …
S Moon, Y Park, DW Ko, IH Suh - International Journal of …, 2016 - journals.sagepub.com
… full-bodyposes in realtime by combining the initially-estimated multiple 3D bodyposes. … , we proposed a real-time motion-tracking technique utilizing multipleKinect sensors. In this …
LJS Silva, DLS da Silva, AB Raposo, L Velho… - Computers & …, 2019 - Elsevier
… results for 2D multi-person poseestimationusingmulti-stage … the 2D humanpose, ie, the anatomical keypoints or body “parts” … optical flow and a Kalman filter to smooth the movement. …
… for estimating 3D humanposes that works in real-time in an … two stages: 2D poseestimation using deep neural networks … first stage in our pipeline is to filter out outliers in the previous …
… Kinect was evaluated in terms of detecting the humanbody … , the real-time motion of the full-bodyposition and orientation … filter [39] and tracking only the legs to locate the real-time foot …
… five challenging actions by building a real-time demo system. It … To filter out the unreliable Kinect joints, we proposed treating … frame showing an unreasonable humanpose. The wrong …
… images captured by multiple RGB-D (two Kinect) cameras, … -filtered method, and body orientation estimationusing principal … Real-timehumanpose recognition in parts from single depth …