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 1.24.3	Download
ONNX	w8a16_mixed_fp16	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a16_mixed_fp16	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® 8 Elite Gen 5 Mobile	167.029 ms	8 - 2259 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Elite Mobile	184.398 ms	1 - 2182 MB	NPU
CVT	ONNX	float	Snapdragon® X2 Elite	176.591 ms	20 - 20 MB	NPU
CVT	ONNX	float	Snapdragon® X Elite	278.481 ms	18 - 18 MB	NPU
CVT	ONNX	float	Snapdragon® X Elite	278.481 ms	18 - 18 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Gen 3 Mobile	219.232 ms	8 - 2573 MB	NPU
CVT	ONNX	float	Qualcomm® QCS8550 (Proxy)	324.375 ms	0 - 24 MB	NPU
CVT	ONNX	float	Qualcomm® QCS9075	317.491 ms	7 - 10 MB	NPU
CVT	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	184.398 ms	1 - 2182 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	231.104 ms	6 - 1303 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Mobile	259.673 ms	3 - 1132 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® X2 Elite	221.752 ms	18 - 18 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® X Elite	360.759 ms	15 - 15 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® X Elite	360.759 ms	15 - 15 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	303.763 ms	6 - 1626 MB	NPU
CVT	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	259.673 ms	3 - 1132 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	179.253 ms	7 - 2297 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Elite Mobile	183.575 ms	7 - 2297 MB	NPU
CVT	QNN_DLC	float	Snapdragon® X2 Elite	181.804 ms	7 - 7 MB	NPU
CVT	QNN_DLC	float	Snapdragon® X Elite	281.634 ms	7 - 7 MB	NPU
CVT	QNN_DLC	float	Snapdragon® X Elite	281.634 ms	7 - 7 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	228.567 ms	7 - 2576 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	430.113 ms	2 - 2214 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	323.682 ms	8 - 9 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8775P	292.561 ms	2 - 2214 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8775P	292.561 ms	2 - 2214 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8775P	292.561 ms	2 - 2214 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS9075	322.826 ms	7 - 17 MB	NPU
CVT	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	440.767 ms	5 - 2718 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA7255P	430.113 ms	2 - 2214 MB	NPU
CVT	QNN_DLC	float	Qualcomm® SA8295P	332.305 ms	1 - 2325 MB	NPU
CVT	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	183.575 ms	7 - 2297 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	193.385 ms	4 - 2349 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite Mobile	211.645 ms	4 - 2330 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X2 Elite	206.64 ms	4 - 4 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X Elite	319.969 ms	4 - 4 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X Elite	319.969 ms	4 - 4 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	246.948 ms	8 - 2594 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS8275 (Proxy)	472.882 ms	2 - 2319 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS8550 (Proxy)	354.734 ms	4 - 1279 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8775P	323.12 ms	2 - 2319 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8775P	323.12 ms	2 - 2319 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8775P	323.12 ms	2 - 2319 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS9075	396.587 ms	4 - 9 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA7255P	472.882 ms	2 - 2319 MB	NPU
CVT	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	211.645 ms	4 - 2330 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	166.523 ms	0 - 2304 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Elite Mobile	182.509 ms	0 - 2353 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	216.769 ms	0 - 2674 MB	NPU
CVT	TFLITE	float	Qualcomm® QCS8550 (Proxy)	299.412 ms	0 - 4 MB	NPU
CVT	TFLITE	float	Qualcomm® QCS9075	317.618 ms	0 - 36 MB	NPU
CVT	TFLITE	float	Qualcomm® QCS8450 (Proxy)	423.956 ms	0 - 2726 MB	NPU
CVT	TFLITE	float	Qualcomm® SA8295P	329.079 ms	0 - 2307 MB	NPU
CVT	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	182.509 ms	0 - 2353 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.