--- library_name: pytorch license: other tags: - bu_auto - android pipeline_tag: other --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/bevdet/web-assets/model_demo.png) # 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](https://github.com/HuangJunJie2017/BEVDet/). This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/bevdet) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary). Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) 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 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/bevdet/releases/v0.51.0/bevdet-onnx-float.zip) | ONNX | w8a16_mixed_fp16 | Universal | ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/bevdet/releases/v0.51.0/bevdet-onnx-w8a16_mixed_fp16.zip) | TFLITE | float | Universal | | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/bevdet/releases/v0.51.0/bevdet-tflite-float.zip) For more device-specific assets and performance metrics, visit **[BEVDet on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/bevdet)**. ### Option 2: Export with Custom Configurations Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/bevdet) 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](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/bevdet) 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® 8 Elite Gen 5 Mobile | 1326.889 ms | 252 - 263 MB | CPU | BEVDet | ONNX | float | Snapdragon® X2 Elite | 606.835 ms | 736 - 736 MB | CPU | BEVDet | ONNX | float | Snapdragon® X Elite | 621.617 ms | 732 - 732 MB | CPU | BEVDet | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 2090.336 ms | 216 - 226 MB | CPU | BEVDet | ONNX | float | Qualcomm® QCS8550 (Proxy) | 2652.368 ms | 187 - 189 MB | CPU | BEVDet | ONNX | float | Qualcomm® QCS9075 | 1523.403 ms | 236 - 250 MB | CPU | BEVDet | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 1381.605 ms | 248 - 262 MB | CPU | BEVDet | ONNX | w8a16_mixed_fp16 | Snapdragon® 8 Elite Gen 5 Mobile | 1565.626 ms | 319 - 333 MB | CPU | BEVDet | ONNX | w8a16_mixed_fp16 | Snapdragon® X2 Elite | 905.903 ms | 708 - 708 MB | CPU | BEVDet | ONNX | w8a16_mixed_fp16 | Snapdragon® X Elite | 889.641 ms | 1239 - 1239 MB | CPU | BEVDet | ONNX | w8a16_mixed_fp16 | Snapdragon® 8 Gen 3 Mobile | 2280.19 ms | 367 - 378 MB | CPU | BEVDet | ONNX | w8a16_mixed_fp16 | Qualcomm® QCS8550 (Proxy) | 2676.944 ms | 398 - 401 MB | CPU | BEVDet | ONNX | w8a16_mixed_fp16 | Qualcomm® QCS9075 | 1855.02 ms | 425 - 435 MB | CPU | BEVDet | ONNX | w8a16_mixed_fp16 | Snapdragon® 8 Elite For Galaxy Mobile | 1535.849 ms | 329 - 343 MB | CPU | BEVDet | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1173.851 ms | 87 - 97 MB | CPU | BEVDet | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 1836.121 ms | 106 - 118 MB | CPU | BEVDet | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 3146.339 ms | 128 - 136 MB | CPU | BEVDet | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 1999.129 ms | 128 - 131 MB | CPU | BEVDet | TFLITE | float | Qualcomm® SA8775P | 2489.569 ms | 127 - 138 MB | CPU | BEVDet | TFLITE | float | Qualcomm® QCS9075 | 2391.956 ms | 126 - 1330 MB | CPU | BEVDet | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 2625.93 ms | 122 - 139 MB | CPU | BEVDet | TFLITE | float | Qualcomm® SA7255P | 3146.339 ms | 128 - 136 MB | CPU | BEVDet | TFLITE | float | Qualcomm® SA8295P | 1789.726 ms | 127 - 137 MB | CPU | BEVDet | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 1275.134 ms | 108 - 121 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](https://arxiv.org/abs/2112.11790) * [Source Model Implementation](https://github.com/HuangJunJie2017/BEVDet/) ## Community * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI. * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).