Rethinking Atrous Convolution for Semantic Image Segmentation
Paper • 1706.05587 • Published
DeepLabV3-Plus-MobileNet: Optimized for Qualcomm Devices
DeepLabV3 is designed for semantic segmentation at multiple scales, trained on the various datasets. It uses MobileNet as a backbone.
This is based on the implementation of DeepLabV3-Plus-MobileNet 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 2.42, ONNX Runtime 1.24.3 | Download |
| ONNX | w8a16 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | Download |
| ONNX | w8a8 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | Download |
| QNN_DLC | float | Universal | QAIRT 2.45 | Download |
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | Download |
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | Download |
| TFLITE | float | Universal | QAIRT 2.45 | Download |
| TFLITE | w8a8 | Universal | QAIRT 2.45 | Download |
For more device-specific assets and performance metrics, visit DeepLabV3-Plus-MobileNet 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 DeepLabV3-Plus-MobileNet on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.semantic_segmentation
Model Stats:
- Model checkpoint: VOC2012
- Input resolution: 513x513
- Number of output classes: 21
- Number of parameters: 5.80M
- Model size (float): 22.2 MB
- Model size (w8a16): 6.67 MB
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| DeepLabV3-Plus-MobileNet | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 4.302 ms | 2 - 176 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | float | Snapdragon® X2 Elite | 5.394 ms | 10 - 10 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | float | Snapdragon® X Elite | 11.053 ms | 10 - 10 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 7.19 ms | 4 - 211 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | float | Qualcomm® QCS8550 (Proxy) | 10.361 ms | 3 - 131 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | float | Qualcomm® QCS9075 | 18.058 ms | 3 - 6 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 5.893 ms | 2 - 170 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 2.962 ms | 0 - 207 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Snapdragon® X2 Elite | 3.711 ms | 7 - 7 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Snapdragon® X Elite | 7.659 ms | 5 - 5 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 5.092 ms | 2 - 220 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Qualcomm® QCS6490 | 1223.297 ms | 89 - 92 MB | CPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Qualcomm® QCS8550 (Proxy) | 7.103 ms | 2 - 158 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Qualcomm® QCS9075 | 8.613 ms | 2 - 4 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Qualcomm® QCM6690 | 588.21 ms | 115 - 123 MB | CPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 3.934 ms | 0 - 185 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 586.218 ms | 93 - 101 MB | CPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 1.463 ms | 0 - 179 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a8 | Snapdragon® X2 Elite | 1.667 ms | 5 - 5 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a8 | Snapdragon® X Elite | 3.921 ms | 5 - 5 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 2.375 ms | 0 - 202 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 3.524 ms | 0 - 9 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a8 | Qualcomm® QCS9075 | 4.464 ms | 1 - 4 MB | NPU |
| DeepLabV3-Plus-MobileNet | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.907 ms | 0 - 176 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 4.307 ms | 3 - 184 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Snapdragon® X2 Elite | 5.601 ms | 3 - 3 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Snapdragon® X Elite | 12.287 ms | 3 - 3 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 8.069 ms | 0 - 206 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 58.629 ms | 0 - 166 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 11.555 ms | 0 - 9 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Qualcomm® SA8775P | 17.509 ms | 1 - 167 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Qualcomm® QCS9075 | 20.059 ms | 3 - 8 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 19.783 ms | 1 - 208 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Qualcomm® SA7255P | 58.629 ms | 0 - 166 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Qualcomm® SA8295P | 19.622 ms | 0 - 170 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 6.284 ms | 0 - 176 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 3.074 ms | 0 - 194 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 4.25 ms | 2 - 2 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Snapdragon® X Elite | 9.114 ms | 2 - 2 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 6.126 ms | 2 - 213 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® QCS6490 | 33.499 ms | 3 - 7 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® QCS8275 (Proxy) | 21.969 ms | 2 - 182 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® QCS8550 (Proxy) | 8.395 ms | 2 - 3 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® SA8775P | 8.901 ms | 2 - 184 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® QCS9075 | 9.733 ms | 3 - 6 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® QCM6690 | 111.918 ms | 2 - 232 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® QCS8450 (Proxy) | 12.139 ms | 2 - 215 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® SA7255P | 21.969 ms | 2 - 182 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Qualcomm® SA8295P | 13.707 ms | 2 - 197 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 4.365 ms | 2 - 178 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 11.976 ms | 2 - 181 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 1.612 ms | 1 - 176 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 2.075 ms | 1 - 1 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Snapdragon® X Elite | 4.63 ms | 1 - 1 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 2.915 ms | 0 - 198 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 15.416 ms | 0 - 3 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® QCS8275 (Proxy) | 10.819 ms | 1 - 170 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 4.206 ms | 1 - 22 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® SA8775P | 4.696 ms | 1 - 172 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 5.128 ms | 0 - 3 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 48.971 ms | 1 - 202 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® QCS8450 (Proxy) | 6.75 ms | 1 - 197 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® SA7255P | 10.819 ms | 1 - 170 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Qualcomm® SA8295P | 6.423 ms | 1 - 168 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 2.185 ms | 1 - 173 MB | NPU |
| DeepLabV3-Plus-MobileNet | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 6.259 ms | 1 - 176 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 4.312 ms | 0 - 183 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 8.09 ms | 0 - 215 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 58.638 ms | 0 - 168 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 11.479 ms | 0 - 2 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Qualcomm® SA8775P | 17.543 ms | 0 - 168 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Qualcomm® QCS9075 | 19.566 ms | 0 - 18 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 19.887 ms | 0 - 212 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Qualcomm® SA7255P | 58.638 ms | 0 - 168 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Qualcomm® SA8295P | 19.618 ms | 0 - 170 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 6.281 ms | 0 - 177 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 1.585 ms | 0 - 183 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 2.942 ms | 0 - 200 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® QCS6490 | 15.911 ms | 0 - 11 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® QCS8275 (Proxy) | 10.839 ms | 0 - 174 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 4.355 ms | 0 - 102 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® SA8775P | 4.622 ms | 0 - 176 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® QCS9075 | 5.114 ms | 0 - 9 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® QCM6690 | 47.33 ms | 0 - 196 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 5.745 ms | 0 - 203 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® SA7255P | 10.839 ms | 0 - 174 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Qualcomm® SA8295P | 6.255 ms | 0 - 172 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 2.081 ms | 0 - 174 MB | NPU |
| DeepLabV3-Plus-MobileNet | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 6.363 ms | 0 - 178 MB | NPU |
License
- The license for the original implementation of DeepLabV3-Plus-MobileNet 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.