Hugging Face's logo

qualcomm
/
BEVFormer

Other
PyTorch
bu_auto
android
Model card Files
xet
 Community
BEVFormer / README.md
qaihm-bot's picture
qaihm-bot
v0.64.0
34b43b9 verified
preview code
|
Raw History Blame Contribute Delete
6.19 kB
metadata 
library_name: pytorch
license: apache-2.0
tags:
  - bu_auto
  - android
pipeline_tag: other

BEVFormer: Optimized for Qualcomm Devices

Bevformer is a SOTA model of interest to the Auto BU.

This is based on the implementation of BEVFormer 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 2.50, ONNX Runtime 1.30.0 Download
QNN_DLC float Universal QAIRT 2.50 Download

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

Model Details

Model Type: Model_use_case.driver_assistance

Model Stats:

  • Input preprocessing: Images must be ImageNet mean/std normalized before inference; the model does not normalize internally.
  • Input resolution: 6 x 3 x 480 x 800
  • Model checkpoint: bevformer_tiny_deformable_optimized_exp_86_epoch_24.pth
  • Model size: 120 MB
  • Number of parameters: 27M

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
BEVFormer ONNX float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 898.972 ms 7 - 786 MB NPU
BEVFormer ONNX float Snapdragon® 8 Elite For Galaxy Mobile 1111.845 ms 3 - 739 MB NPU
BEVFormer ONNX float Snapdragon® X2 Elite 887.121 ms 29 - 29 MB NPU
BEVFormer ONNX float Snapdragon® X Elite 1707.571 ms 50 - 50 MB NPU
BEVFormer ONNX float Snapdragon® 8 Gen 3 Mobile 1354.737 ms 0 - 972 MB NPU
BEVFormer ONNX float Snapdragon® 8 Gen 1 Mobile 1769.929 ms 29 - 986 MB NPU
BEVFormer ONNX float Qualcomm® Dragonwing™ IQ-8275 1920.481 ms 29 - 62 MB NPU
BEVFormer ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 1642.585 ms 0 - 52 MB NPU
BEVFormer ONNX float Qualcomm® QCS8450 1769.929 ms 29 - 986 MB NPU
BEVFormer ONNX float Qualcomm® Dragonwing™ IQ-9075 1827.543 ms 29 - 61 MB NPU
BEVFormer ONNX float Qualcomm® Dragonwing™ IQ-X7181 1707.571 ms 50 - 50 MB NPU
BEVFormer ONNX float Qualcomm® Dragonwing™ Q-8750 1111.845 ms 3 - 739 MB NPU
BEVFormer QNN_DLC float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 880.995 ms 20 - 860 MB NPU
BEVFormer QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 1438.781 ms 10 - 801 MB NPU
BEVFormer QNN_DLC float Snapdragon® X2 Elite 924.225 ms 29 - 29 MB NPU
BEVFormer QNN_DLC float Snapdragon® X Elite 1703.876 ms 29 - 29 MB NPU
BEVFormer QNN_DLC float Snapdragon® 8 Gen 3 Mobile 1342.843 ms 27 - 1002 MB NPU
BEVFormer QNN_DLC float Snapdragon® 8 Gen 1 Mobile 1415.097 ms 28 - 1013 MB NPU
BEVFormer QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 1996.15 ms 29 - 63 MB NPU
BEVFormer QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 1584.595 ms 29 - 32 MB NPU
BEVFormer QNN_DLC float Qualcomm® SA8650P 2011.958 ms 26 - 863 MB NPU
BEVFormer QNN_DLC float Qualcomm® SA8255P 2011.958 ms 26 - 863 MB NPU
BEVFormer QNN_DLC float Qualcomm® QCS8450 1415.097 ms 28 - 1013 MB NPU
BEVFormer QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 1789.621 ms 29 - 62 MB NPU
BEVFormer QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 1703.876 ms 29 - 29 MB NPU
BEVFormer QNN_DLC float Qualcomm® Dragonwing™ Q-8750 1438.781 ms 10 - 801 MB NPU
BEVFormer QNN_DLC float Qualcomm® SA8295P 1739.194 ms 26 - 720 MB NPU

License

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

References

Community