サインイン
インサイト - Pose Estimation
Accurate Real-time Relative Pose Estimation from Triple Point-line Images
Decoupled rotation and translation estimation improves accuracy in three-view pose estimation.
GatedUniPose: An Efficient and Robust Approach for Human Pose Estimation Combining UniRepLKNet and Gated Convolution
GatedUniPose, a novel pose estimation method, combines UniRepLKNet and Gated Convolution to achieve significant performance improvements in handling complex scenes and occlusion challenges, while maintaining a relatively small number of parameters.
GateAttentionPose: An Efficient and Accurate Approach for Human Pose Estimation with Agent Attention and Improved Gated Convolutions
GateAttentionPose is an innovative framework that enhances both accuracy and computational efficiency for human pose estimation tasks by introducing the Agent Attention module and the Gate-Enhanced Feedforward Block (GEFB).
