CVT: Optimized for Qualcomm Devices

Cross-View Transformer generates real-time bird's-eye view maps from multiple vehicle cameras for autonomous driving.

This is based on the implementation of CVT 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 CVT 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 CVT on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.driver_assistance

Model Stats:

• Model checkpoint: vehicles_50k.pt
• Inference latency: RealTime
• Input resolution: 1x6x3x224x480
• Number of parameters: 1.33M
• Model size (float): 5.18 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
CVT	ONNX	float	Snapdragon® X2 Elite	182.88 ms	8 - 8 MB	NPU
CVT	ONNX	float	Snapdragon® X Elite	278.255 ms	20 - 20 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Gen 3 Mobile	230.186 ms	8 - 2694 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Gen 1 Mobile	436.855 ms	8 - 2781 MB	NPU
CVT	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	326.174 ms	0 - 25 MB	NPU
CVT	ONNX	float	Qualcomm® QCS8450	436.855 ms	8 - 2781 MB	NPU
CVT	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	278.255 ms	20 - 20 MB	NPU
CVT	ONNX	float	Qualcomm® Dragonwing™ Q-8750	185.751 ms	3 - 2380 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Elite Mobile	185.751 ms	3 - 2380 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	168.937 ms	3 - 2415 MB	NPU
CVT	QNN_DLC	float	Snapdragon® X2 Elite	183.244 ms	7 - 7 MB	NPU
CVT	QNN_DLC	float	Snapdragon® X Elite	282.174 ms	7 - 7 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	222.557 ms	0 - 2568 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	437.994 ms	7 - 2720 MB	NPU
CVT	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	424.976 ms	2 - 2214 MB	NPU
CVT	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	335.702 ms	8 - 12 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8775P	292.759 ms	1 - 2213 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8650P	292.759 ms	1 - 2213 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8255P	292.759 ms	1 - 2213 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS8450	437.994 ms	7 - 2720 MB	NPU
CVT	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	282.277 ms	7 - 17 MB	NPU
CVT	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	282.174 ms	7 - 7 MB	NPU
CVT	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	183.886 ms	7 - 2297 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA7255P	424.976 ms	2 - 2214 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8295P	334.683 ms	0 - 2324 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Elite Mobile	183.886 ms	7 - 2297 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	169.423 ms	7 - 2318 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	221.677 ms	1 - 2674 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	405.545 ms	1 - 2726 MB	NPU
CVT	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	319.534 ms	0 - 4 MB	NPU
CVT	TFLITE	float	Qualcomm® QCS8450	405.545 ms	1 - 2726 MB	NPU
CVT	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	182.745 ms	1 - 2364 MB	NPU
CVT	TFLITE	float	Qualcomm® SA8295P	332.107 ms	0 - 2306 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Elite Mobile	182.745 ms	1 - 2364 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	168.2 ms	0 - 2315 MB	NPU

License

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

References

• Cross-view Transformers for real-time Map-view Semantic Segmentation
• 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.