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 2.42, ONNX Runtime 1.24.1	Download
ONNX	w8a16_mixed_fp16	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.17.0	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® X Elite	287.315 ms	18 - 18 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Gen 3 Mobile	227.134 ms	8 - 2577 MB	NPU
CVT	ONNX	float	Qualcomm® QCS8550 (Proxy)	336.234 ms	0 - 26 MB	NPU
CVT	ONNX	float	Qualcomm® QCS9075	327.451 ms	7 - 10 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	189.979 ms	4 - 2222 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	173.494 ms	8 - 2259 MB	NPU
CVT	ONNX	float	Snapdragon® X2 Elite	186.137 ms	20 - 20 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® X Elite	365.903 ms	16 - 16 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	303.459 ms	15 - 1609 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Qualcomm® QCS8550 (Proxy)	400.69 ms	13 - 2070 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	263.913 ms	12 - 1133 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	217.984 ms	16 - 1268 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® X2 Elite	221.965 ms	18 - 18 MB	NPU
CVT	QNN_DLC	float	Snapdragon® X Elite	277.336 ms	7 - 7 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	224.075 ms	4 - 2551 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	420.883 ms	0 - 2185 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	334.853 ms	8 - 10 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8775P	291.401 ms	1 - 2187 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	426.183 ms	7 - 2650 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA7255P	420.883 ms	0 - 2185 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8295P	331.888 ms	0 - 2239 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	183.305 ms	7 - 2172 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	166.578 ms	7 - 2258 MB	NPU
CVT	QNN_DLC	float	Snapdragon® X2 Elite	179.712 ms	7 - 7 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	226.653 ms	0 - 2563 MB	NPU
CVT	TFLITE	float	Qualcomm® QCS8275 (Proxy)	426.132 ms	0 - 2194 MB	NPU
CVT	TFLITE	float	Qualcomm® QCS8550 (Proxy)	333.195 ms	0 - 4 MB	NPU
CVT	TFLITE	float	Qualcomm® SA8775P	931.204 ms	0 - 2193 MB	NPU
CVT	TFLITE	float	Qualcomm® QCS8450 (Proxy)	456.96 ms	1 - 2731 MB	NPU
CVT	TFLITE	float	Qualcomm® SA7255P	426.132 ms	0 - 2194 MB	NPU
CVT	TFLITE	float	Qualcomm® SA8295P	337.862 ms	0 - 2290 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	188.748 ms	0 - 2243 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	182.012 ms	0 - 2288 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.