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
ESS DNN Stereo Disparity
Model
NVIDIA
NVIDIA
ESS DNN Stereo Disparity

ESS is a DNN that estimates disparity for a stereo image pair and returns a continuous disparity map for the given left image.

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):Building depth, localization, mapping, and navigation capabilities for robotics and autonomous machines.
Model Type:Depth estimation.
Intended Users:Developers building and/or customizing robotics applications
Output:Continuous Disparity Map
Describe how the model works:Predicted continuous disparity map generated from left and right image tensors.
Technical Limitations:Prediction accuracy varies with pixels that are whiter, transparent, and reflective in nature; for sharp boundaries and fine edges; and for flat featureless areas. Visual artifacts can appear in non-overlapping regions and from overfitting.
Verified to have met prescribed NVIDIA standards?Yes.
Performance Metrics:Bad Pixels (Bad Pixels) and Mean Average Error (MAE); Bad Pixel Percentage > 2px.
Potential Known Risks:Imprecise disparity and distance predictions can lead to inability to localize robot and lead to collisions.
Licensing:https://developer.download.nvidia.com/licenses/tao_toolkit_21-08_models_eula.pdf