BEVDet: Optimized for Qualcomm Devices

BEVDet is a machine learning model for generating a birds eye view represenation from the sensors(cameras) mounted on a vehicle.

This is based on the implementation of BEVDet 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	ONNX Runtime 1.24.1	Download
ONNX	w8a16_mixed_fp16	Universal	ONNX Runtime 1.24.1	Download
TFLITE	float	Universal	TFLite 2.17.0	Download

For more device-specific assets and performance metrics, visit BEVDet 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 BEVDet on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.driver_assistance

Model Stats:

• Model checkpoint: bevdet-r50.pth
• Input resolution: 1 x 6 x 3 x 256 x 704
• Number of parameters: 44M
• Model size: 171 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
BEVDet	ONNX	float	Snapdragon® X2 Elite	597.959 ms	732 - 732 MB	CPU
BEVDet	ONNX	float	Snapdragon® X Elite	668.299 ms	732 - 732 MB	CPU
BEVDet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	2414.581 ms	211 - 221 MB	CPU
BEVDet	ONNX	float	Qualcomm® QCS8550 (Proxy)	2627.235 ms	187 - 189 MB	CPU
BEVDet	ONNX	float	Qualcomm® QCS9075	1513.66 ms	240 - 257 MB	CPU
BEVDet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	1335.645 ms	245 - 258 MB	CPU
BEVDet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1315.712 ms	249 - 259 MB	CPU
BEVDet	ONNX	w8a16_mixed_fp16	Snapdragon® X2 Elite	897.878 ms	713 - 713 MB	CPU
BEVDet	ONNX	w8a16_mixed_fp16	Snapdragon® X Elite	1005.14 ms	712 - 712 MB	CPU
BEVDet	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	2345.819 ms	361 - 371 MB	CPU
BEVDet	ONNX	w8a16_mixed_fp16	Qualcomm® QCS8550 (Proxy)	2732.696 ms	397 - 403 MB	CPU
BEVDet	ONNX	w8a16_mixed_fp16	Qualcomm® QCS9075	1869.644 ms	423 - 430 MB	CPU
BEVDet	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	1571.022 ms	321 - 331 MB	CPU
BEVDet	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	1854.694 ms	321 - 335 MB	CPU
BEVDet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	1633.888 ms	103 - 115 MB	CPU
BEVDet	TFLITE	float	Qualcomm® QCS8275 (Proxy)	3146.061 ms	128 - 138 MB	CPU
BEVDet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	2070.354 ms	127 - 130 MB	CPU
BEVDet	TFLITE	float	Qualcomm® SA8775P	2527.606 ms	129 - 140 MB	CPU
BEVDet	TFLITE	float	Qualcomm® QCS9075	2423.928 ms	127 - 1474 MB	CPU
BEVDet	TFLITE	float	Qualcomm® QCS8450 (Proxy)	2620.623 ms	129 - 148 MB	CPU
BEVDet	TFLITE	float	Qualcomm® SA7255P	3146.061 ms	128 - 138 MB	CPU
BEVDet	TFLITE	float	Qualcomm® SA8295P	1925.185 ms	78 - 85 MB	CPU
BEVDet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	1285.09 ms	105 - 119 MB	CPU
BEVDet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1049.079 ms	88 - 99 MB	CPU

License

• The license for the original implementation of BEVDet can be found [here](https://github.com/HuangJunJie2017/BEVDet/blob/dev3.0/LICENSE https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf).

References

• BEVDet: High-Performance Multi-Camera 3D Object Detection in Bird-Eye-View
• 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.