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
FoundationStereo
Model
NVIDIA
NVIDIA
FoundationStereo

stereo depth estimation model.

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):Depth Perception for Robotics, Autonomous Vision Navigation and Space Mapping
Model Type:Zero-shot Depth Estimation
Intended Users:This model is intended for uses who need accurate 3D depth for 2D Stereo Vision
Output:Disparity or Depth Map
Describe how the model works:1. Pass in two Stereo RGB images. 2. Run the model on the stereo input. 3. Save the disparity map and use camera intrinsic valus to obtain the corresponding depth map.
Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless ofNot Applicable
Technical Limitations:This model may not work well for detecting partially or fully occluded objects.
Verified to have met prescribed NVIDIA standards:Yes
Performance Metrics:Percentage of bad pixels, End-point Error (EPE)
Potential Known RisksPoor depth perception from a failure case could lead to hallucination of objects into the depth scene or wrong depth estimation of objects in a scene
Licensing:NVIDIA Open Model License