OpenPose: Pose Estimation
OpenPose, an open-source real-time multi-person pose estimation framework developed by Carnegie Mellon University, detects human body, hand, and facial key points. Using CNN and Part Affinity Fields (PAFs), it locates and associates joints in multi-person scenarios, widely applied in action recognition, motion analysis, human-computer interaction, and AR. Its strength lies in high-accuracy multi-person tracking and cross-platform compatibility (CPU/GPU), though resource-intensive for edge deployment. Challenges include occlusion handling, model optimization, and robustness in low-light conditions.
Source Model
- Input shape: 1x3x224xx224
- Number of parameters: 49.89M
- Model size: 200.04M
- Output shape: 1x38x28x28, 1x19x28x28
The source model can be found here
Performance Reference
Please search model by model name in Model Farm
Inference & Model Conversion
Please search model by model name in Model Farm
License
Source Model: OPENPOSE_LICENSE
Deployable Model: OPENPOSE_LICENSE