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 2.45, ONNX Runtime 1.27.1	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:

• 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® X2 Elite	205.687 ms	5 - 5 MB	NPU
StereoNet	ONNX	float	Snapdragon® X Elite	380.583 ms	44 - 44 MB	NPU
StereoNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	299.99 ms	7 - 4398 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	491.744 ms	3 - 9 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	428.175 ms	0 - 49 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	534.614 ms	3 - 9 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	380.583 ms	44 - 44 MB	NPU
StereoNet	ONNX	float	Qualcomm® Dragonwing™ Q-8750	251.198 ms	3 - 3234 MB	NPU
StereoNet	ONNX	float	Snapdragon® 8 Elite Mobile	251.198 ms	3 - 3234 MB	NPU
StereoNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	202.336 ms	3 - 3299 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® X2 Elite	192.792 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® X Elite	362.642 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	287.9 ms	3 - 4450 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	483.015 ms	3 - 10 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	1293.86 ms	0 - 3260 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	435.821 ms	3 - 7 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8775P	462.152 ms	2 - 3261 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8650P	462.152 ms	2 - 3261 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8255P	462.152 ms	2 - 3261 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	448.457 ms	3 - 9 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	362.642 ms	3 - 3 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	238.33 ms	1 - 3242 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA7255P	1293.86 ms	0 - 3260 MB	NPU
StereoNet	QNN_DLC	float	Qualcomm® SA8295P	516.045 ms	0 - 3366 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Elite Mobile	238.33 ms	1 - 3242 MB	NPU
StereoNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	187.034 ms	3 - 3301 MB	NPU
StereoNet	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	279.882 ms	51 - 3751 MB	NPU
StereoNet	TFLITE	float	Snapdragon® 8 Elite Mobile	279.882 ms	51 - 3751 MB	NPU
StereoNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	267.667 ms	52 - 3842 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.