Objects as Points
Paper • 1904.07850 • Published
CenterNet-2D: Optimized for Qualcomm Devices
CenterNet-2D is machine learning model that detects objects by finding their center points.
This is based on the implementation of CenterNet-2D 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 |
|---|---|---|---|---|
| PRECOMPILED_QNN_ONNX | float | Snapdragon® X2 Elite | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® X Elite | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 1 Mobile | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS8550 (Proxy) | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite Mobile | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS9075 | QAIRT 2.45, ONNX Runtime 1.25.0 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® X2 Elite | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® X Elite | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 3 Mobile | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 1 Mobile | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® QCS8550 (Proxy) | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® SA8775P | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite Mobile | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite Gen 5 Mobile | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® SA7255P | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® SA8295P | QAIRT 2.45 | Download |
| QNN_CONTEXT_BINARY | float | Qualcomm® QCS9075 | QAIRT 2.45 | Download |
For more device-specific assets and performance metrics, visit CenterNet-2D 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 CenterNet-2D on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.object_detection
Model Stats:
- Model checkpoint: ctdet_coco_dla_2x.pth
- Input resolution: 1 x 3 x 512 x 512
- Number of parameters: 20.2M
- Model size: 37.6 MB
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Snapdragon® X2 Elite | 237.548 ms | 36 - 36 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Snapdragon® X Elite | 362.241 ms | 56 - 56 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 3 Mobile | 264.551 ms | 19 - 25 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Gen 1 Mobile | 729.164 ms | 10 - 24 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS8550 (Proxy) | 357.738 ms | 0 - 62 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS8450 | 729.164 ms | 10 - 24 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 231.507 ms | 15 - 23 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS9075 | 365.382 ms | 8 - 14 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Snapdragon® 8 Elite Mobile | 289.087 ms | 14 - 21 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS8750 | 289.087 ms | 14 - 21 MB | NPU |
| CenterNet-2D | PRECOMPILED_QNN_ONNX | float | Qualcomm® QCS7181 | 362.241 ms | 56 - 56 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Snapdragon® X2 Elite | 236.75 ms | 3 - 3 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Snapdragon® X Elite | 358.474 ms | 3 - 3 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 3 Mobile | 269.648 ms | 3 - 11 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Gen 1 Mobile | 748.724 ms | 3 - 18 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® QCS8275 | 516.832 ms | 1 - 9 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® QCS8550 (Proxy) | 356.268 ms | 3 - 5 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® QCS8450 | 748.724 ms | 3 - 18 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite Gen 5 Mobile | 231.06 ms | 3 - 13 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® SA7255P | 516.832 ms | 1 - 9 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® QCS9075 | 365.638 ms | 5 - 15 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Snapdragon® 8 Elite Mobile | 289.056 ms | 1 - 10 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® SA8295P | 442.26 ms | 0 - 6 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® QCS8750 | 289.056 ms | 1 - 10 MB | NPU |
| CenterNet-2D | QNN_CONTEXT_BINARY | float | Qualcomm® QCS7181 | 358.474 ms | 3 - 3 MB | NPU |
License
- The license for the original implementation of CenterNet-2D can be found here.
References
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.