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 2.50, ONNX Runtime 1.30.0	Download
QNN_DLC	float	Universal	QAIRT 2.50	Download
TFLITE	float	Universal	QAIRT 2.50	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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
StereoNet	ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	204.555 ms	5 - 2837 MB	NPU
StereoNet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	244.837 ms	3 - 2785 MB	NPU
StereoNet	ONNX	float	Snapdragon® X2 Elite	206.046 ms	5 - 5 MB	NPU
StereoNet	ONNX	float	Snapdragon® X Elite	366.124 ms	39 - 39 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	474.788 ms	3 - 10 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	404.773 ms	5 - 8 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	443.626 ms	3 - 9 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	366.124 ms	39 - 39 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ Q-8750	244.837 ms	3 - 2785 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	187.123 ms	3 - 2846 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	226.999 ms	0 - 2784 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® X2 Elite	188.883 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® X Elite	346.073 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	270.213 ms	3 - 3860 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	457.985 ms	3 - 10 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	418.982 ms	3 - 1083 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8650P	461.966 ms	1 - 3260 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8255P	461.966 ms	1 - 3260 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	425.365 ms	3 - 9 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	346.073 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	226.999 ms	0 - 2784 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8295P	507.265 ms	0 - 2944 MB	NPU
StereoNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	278.885 ms	60 - 3301 MB	NPU
StereoNet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	291.729 ms	73 - 3219 MB	NPU
StereoNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	354.328 ms	73 - 4245 MB	NPU
StereoNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	574.789 ms	74 - 199 MB	NPU
StereoNet	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	568.559 ms	74 - 77 MB	NPU
StereoNet	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	554.801 ms	72 - 196 MB	NPU
StereoNet	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	291.729 ms	73 - 3219 MB	NPU
StereoNet	TFLITE	float	Qualcomm® SA8295P	616.772 ms	74 - 3210 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.