StereoNet: Optimized for Qualcomm Devices

StereoNet is an end-to-end deep architecture for real-time stereo matching that produces high-quality, edge-preserved disparity maps from a rectified stereo image pair.

This is based on the implementation of StereoNet found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime	Precision	Chipset	SDK Versions	Download
ONNX	float	Universal	QAIRT 2.42, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit StereoNet on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

• Custom weights (e.g., fine-tuned checkpoints)
• Custom input shapes
• Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for StereoNet on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.depth_estimation

Model Stats:

• Model checkpoint: KeystoneDepth (epoch=21-step=696366.ckpt)
• Input resolution: 786x490
• Number of parameters: 1.94M
• Model size (float): 7.41 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
StereoNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	195.313 ms	6 - 1354 MB	NPU
StereoNet	ONNX	float	Snapdragon® X Elite	331.391 ms	158 - 158 MB	NPU
StereoNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	263.072 ms	6 - 1988 MB	NPU
StereoNet	ONNX	float	Qualcomm® QCS8550 (Proxy)	333.649 ms	0 - 47 MB	NPU
StereoNet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	218.716 ms	2 - 1320 MB	NPU
StereoNet	ONNX	float	Qualcomm® QCS9075	512.566 ms	3 - 48 MB	NPU
StereoNet	ONNX	float	Qualcomm® QCS8750	218.716 ms	2 - 1320 MB	NPU
StereoNet	ONNX	float	Qualcomm® QCS7181	331.391 ms	158 - 158 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	187.861 ms	5 - 3264 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® X2 Elite	191.232 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® X Elite	366.32 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	286.471 ms	3 - 4454 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® QCS8275	1293.694 ms	1 - 3259 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	395.849 ms	3 - 6 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8775P	462.094 ms	1 - 3260 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8650P	462.094 ms	1 - 3260 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8255P	462.094 ms	1 - 3260 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA7255P	1293.694 ms	1 - 3259 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8295P	516.099 ms	1 - 3367 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	236.109 ms	0 - 3244 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® QCS9075	512.015 ms	3 - 9 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® QCS8750	236.109 ms	0 - 3244 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® QCS7181	366.32 ms	3 - 3 MB	NPU
StereoNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	277.345 ms	72 - 3821 MB	NPU
StereoNet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	274.556 ms	74 - 3773 MB	NPU
StereoNet	TFLITE	float	Qualcomm® QCS9075	663.435 ms	72 - 202 MB	NPU
StereoNet	TFLITE	float	Qualcomm® QCS8750	274.556 ms	74 - 3773 MB	NPU

License

• The license for the original implementation of StereoNet can be found here.

References

• StereoNet: Guided Hierarchical Refinement for Real-Time Edge-Aware Depth Prediction
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.