ContainersNVIDIADeepSeek-R1NVIDIALlama-3.1-Nemotron-70B-InstructNVIDIAPyTorch
CollectionsNVIDIAOmniverse Kit (FB)NVIDIADeepStream SDKNVIDIAOmniverse Kit App Streaming
ModelsNVIDIAStyleGAN3 pretrained modelsNVIDIAPeopleNetNVIDIATrafficCamNet
ResourcesNVIDIARiva Skills Quick Start
Helm ChartsNVIDIAGPU OperatorNVIDIANVIDIA NIM Operator
NVIDIA
NVIDIA
FoundationPose
Model
NVIDIA
NVIDIA
FoundationPose

FoundationPose is a unified foundation model for 6D object pose estimation and tracking of objects.

This model is backed by NVIDIA's Plus Plus (++) Promise
to learn more about the quality of the datasets used to train this model.
FieldResponse
Intended Application(s) & Domain(s):Learning pose and tracking vectors for robotics and autonomous machines.
Model Type:Pose and tracking estimation.
Intended Users:Developers building and/or customizing robotics applications.
Output:Vector of dimension six, containing pose and tracking information.
Describe how the model works:Predicts pose and tracking information of objects when input RGB or CAD is passed into the model.
Technical Limitations:FoundationPose fails to work with transparent objects, glassy materials and acutely blurry images.
Performance Metrics:AR score.
Potential Known Risks:N/A
Licensing:https://developer.download.nvidia.com/licenses/tao_toolkit_21-08_models_eula.pdf