RTMDet: Optimized for Qualcomm Devices
RTMDet is a highly efficient model for real-time object detection,capable of predicting both the bounding boxes and classes of objects within an image.It is highly optimized for real-time applications, making it reliable for industrial and commercial use
This is based on the implementation of RTMDet 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
Due to licensing restrictions, we cannot distribute pre-exported model assets for this model. 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
See our repository for RTMDet on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.object_detection
Model Stats:
- Model checkpoint: RTMDet Medium
- Input resolution: 640x640
- Number of parameters: 27.5M
- Model size (float): 105 MB
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| RTMDet | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 5.977 ms | 5 - 189 MB | NPU |
| RTMDet | ONNX | float | Snapdragon® X2 Elite | 8.176 ms | 53 - 53 MB | NPU |
| RTMDet | ONNX | float | Snapdragon® X Elite | 14.152 ms | 51 - 51 MB | NPU |
| RTMDet | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 10.735 ms | 5 - 236 MB | NPU |
| RTMDet | ONNX | float | Qualcomm® QCS8550 (Proxy) | 13.618 ms | 0 - 54 MB | NPU |
| RTMDet | ONNX | float | Qualcomm® QCS9075 | 23.529 ms | 5 - 12 MB | NPU |
| RTMDet | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 8.301 ms | 3 - 185 MB | NPU |
| RTMDet | ONNX | w8a16_mixed_fp16 | Snapdragon® 8 Elite Gen 5 Mobile | 10.395 ms | 3 - 328 MB | NPU |
| RTMDet | ONNX | w8a16_mixed_fp16 | Snapdragon® X2 Elite | 11.244 ms | 32 - 32 MB | NPU |
| RTMDet | ONNX | w8a16_mixed_fp16 | Snapdragon® X Elite | 29.67 ms | 29 - 29 MB | NPU |
| RTMDet | ONNX | w8a16_mixed_fp16 | Snapdragon® 8 Gen 3 Mobile | 22.111 ms | 3 - 386 MB | NPU |
| RTMDet | ONNX | w8a16_mixed_fp16 | Qualcomm® QCS8550 (Proxy) | 28.29 ms | 2 - 39 MB | NPU |
| RTMDet | ONNX | w8a16_mixed_fp16 | Qualcomm® QCS9075 | 33.104 ms | 2 - 5 MB | NPU |
| RTMDet | ONNX | w8a16_mixed_fp16 | Snapdragon® 8 Elite For Galaxy Mobile | 14.303 ms | 1 - 300 MB | NPU |
| RTMDet | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.82 ms | 0 - 210 MB | NPU |
| RTMDet | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 11.725 ms | 0 - 286 MB | NPU |
| RTMDet | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 83.986 ms | 0 - 207 MB | NPU |
| RTMDet | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 15.862 ms | 0 - 3 MB | NPU |
| RTMDet | TFLITE | float | Qualcomm® SA8775P | 22.952 ms | 0 - 208 MB | NPU |
| RTMDet | TFLITE | float | Qualcomm® QCS9075 | 24.387 ms | 0 - 62 MB | NPU |
| RTMDet | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 37.607 ms | 0 - 347 MB | NPU |
| RTMDet | TFLITE | float | Qualcomm® SA7255P | 83.986 ms | 0 - 207 MB | NPU |
| RTMDet | TFLITE | float | Qualcomm® SA8295P | 29.928 ms | 0 - 268 MB | NPU |
| RTMDet | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 9.146 ms | 0 - 208 MB | NPU |
License
- The license for the original implementation of RTMDet 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.