Segformer-Base: Optimized for Qualcomm Devices
Segformer Base is a machine learning model that predicts masks and classes of objects in an image.
This is based on the implementation of Segformer-Base 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 2.45, ONNX Runtime 1.27.1 | Download |
| ONNX | w8a16 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | Download |
| ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | 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 Segformer-Base 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 Segformer-Base on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.semantic_segmentation
Model Stats:
- Input resolution: 512x512
- Model checkpoint: nvidia/segformer-b0-finetuned-ade-512-512
- Model size (float): 14.4 MB
- Model size (w8a16): 4.57 MB
- Model size (w8a8): 3.90 MB
- Number of output classes: 150
- Number of parameters: 3.75M
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| Segformer-Base | ONNX | float | Snapdragon® X2 Elite | 13.576 ms | 35 - 35 MB | NPU |
| Segformer-Base | ONNX | float | Snapdragon® X Elite | 24.569 ms | 35 - 35 MB | NPU |
| Segformer-Base | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 17.327 ms | 24 - 241 MB | NPU |
| Segformer-Base | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 24.66 ms | 0 - 222 MB | NPU |
| Segformer-Base | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 24.361 ms | 22 - 25 MB | NPU |
| Segformer-Base | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 23.66 ms | 0 - 138 MB | NPU |
| Segformer-Base | ONNX | float | Qualcomm® QCS8450 | 24.66 ms | 0 - 222 MB | NPU |
| Segformer-Base | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 25.656 ms | 23 - 26 MB | NPU |
| Segformer-Base | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 24.569 ms | 35 - 35 MB | NPU |
| Segformer-Base | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 14.417 ms | 20 - 211 MB | NPU |
| Segformer-Base | ONNX | float | Snapdragon® 8 Elite Mobile | 14.417 ms | 20 - 211 MB | NPU |
| Segformer-Base | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 13.269 ms | 21 - 213 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Snapdragon® X2 Elite | 5.599 ms | 18 - 18 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Snapdragon® X Elite | 13.599 ms | 17 - 17 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 9.427 ms | 13 - 243 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Snapdragon® 8 Gen 1 Mobile | 17.796 ms | 11 - 244 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 54.107 ms | 14 - 16 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 13.36 ms | 9 - 15 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® QCS8450 | 17.796 ms | 11 - 244 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 14.482 ms | 13 - 15 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 13.599 ms | 17 - 17 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 92.765 ms | 12 - 240 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 17.112 ms | 13 - 234 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 7.285 ms | 13 - 213 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Snapdragon® 8 Elite Mobile | 7.285 ms | 13 - 213 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 5.669 ms | 13 - 218 MB | NPU |
| Segformer-Base | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 17.112 ms | 13 - 234 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Snapdragon® X2 Elite | 1.927 ms | 9 - 9 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Snapdragon® X Elite | 5.011 ms | 5 - 5 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 3.222 ms | 0 - 213 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 7.095 ms | 0 - 216 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 19.53 ms | 7 - 9 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 4.625 ms | 5 - 9 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 4.757 ms | 5 - 11 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® QCS8450 | 7.095 ms | 0 - 216 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 5.15 ms | 8 - 11 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 5.011 ms | 5 - 5 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 30.773 ms | 7 - 235 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 5.411 ms | 9 - 205 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 2.392 ms | 0 - 190 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 2.392 ms | 0 - 190 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 1.912 ms | 0 - 193 MB | NPU |
| Segformer-Base | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 5.411 ms | 9 - 205 MB | NPU |
| Segformer-Base | QNN_DLC | float | Snapdragon® X2 Elite | 12.515 ms | 3 - 3 MB | NPU |
| Segformer-Base | QNN_DLC | float | Snapdragon® X Elite | 23.562 ms | 3 - 3 MB | NPU |
| Segformer-Base | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 16.284 ms | 3 - 229 MB | NPU |
| Segformer-Base | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 29.599 ms | 3 - 227 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 23.26 ms | 3 - 18 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 48.86 ms | 1 - 190 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 22.478 ms | 3 - 5 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® SA8775P | 24.416 ms | 1 - 193 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® SA8650P | 24.416 ms | 1 - 193 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® SA8255P | 24.416 ms | 1 - 193 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® QCS8450 | 29.599 ms | 3 - 227 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 25.029 ms | 5 - 19 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 23.562 ms | 3 - 3 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 13.194 ms | 0 - 197 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® SA7255P | 48.86 ms | 1 - 190 MB | NPU |
| Segformer-Base | QNN_DLC | float | Qualcomm® SA8295P | 27.83 ms | 0 - 187 MB | NPU |
| Segformer-Base | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 13.194 ms | 0 - 197 MB | NPU |
| Segformer-Base | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 13.66 ms | 3 - 196 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 8.956 ms | 2 - 2 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Snapdragon® X Elite | 19.51 ms | 2 - 2 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 15.413 ms | 2 - 239 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 19.414 ms | 0 - 9 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-8275 | 32.475 ms | 2 - 204 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 20.351 ms | 2 - 3 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® SA8775P | 20.197 ms | 2 - 206 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® SA8650P | 20.197 ms | 2 - 206 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® SA8255P | 20.197 ms | 2 - 206 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 20.551 ms | 1 - 9 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 19.51 ms | 2 - 2 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 108.126 ms | 2 - 256 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 34.055 ms | 2 - 232 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 11.23 ms | 1 - 207 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Qualcomm® SA7255P | 32.475 ms | 2 - 204 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Snapdragon® 8 Elite Mobile | 11.23 ms | 1 - 207 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 9.767 ms | 2 - 217 MB | NPU |
| Segformer-Base | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 34.055 ms | 2 - 232 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 2.57 ms | 1 - 1 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Snapdragon® X Elite | 6.298 ms | 1 - 1 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 3.956 ms | 0 - 196 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 8.639 ms | 0 - 198 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 21.771 ms | 1 - 6 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 5.502 ms | 0 - 5 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 10.836 ms | 1 - 174 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 5.777 ms | 1 - 4 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® SA8775P | 6.391 ms | 1 - 176 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® SA8650P | 6.391 ms | 1 - 176 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® SA8255P | 6.391 ms | 1 - 176 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 8.639 ms | 0 - 198 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 6.149 ms | 1 - 6 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 6.298 ms | 1 - 1 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 37.04 ms | 2 - 189 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 6.842 ms | 0 - 179 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 2.838 ms | 0 - 176 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® SA7255P | 10.836 ms | 1 - 174 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Qualcomm® SA8295P | 7.795 ms | 0 - 172 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 2.838 ms | 0 - 176 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 2.247 ms | 1 - 184 MB | NPU |
| Segformer-Base | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 6.842 ms | 0 - 179 MB | NPU |
| Segformer-Base | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 16.445 ms | 9 - 237 MB | NPU |
| Segformer-Base | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 29.286 ms | 9 - 234 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 23.296 ms | 9 - 32 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 48.967 ms | 10 - 200 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 22.72 ms | 8 - 12 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® SA8775P | 24.485 ms | 9 - 202 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® SA8650P | 24.485 ms | 9 - 202 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® SA8255P | 24.485 ms | 9 - 202 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® QCS8450 | 29.286 ms | 9 - 234 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 29.123 ms | 8 - 31 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 13.394 ms | 8 - 204 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® SA7255P | 48.967 ms | 10 - 200 MB | NPU |
| Segformer-Base | TFLITE | float | Qualcomm® SA8295P | 27.854 ms | 10 - 202 MB | NPU |
| Segformer-Base | TFLITE | float | Snapdragon® 8 Elite Mobile | 13.394 ms | 8 - 204 MB | NPU |
| Segformer-Base | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 13.704 ms | 8 - 199 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 7.092 ms | 1 - 207 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 11.995 ms | 2 - 208 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 134.208 ms | 13 - 48 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 9.58 ms | 2 - 13 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-8275 | 18.273 ms | 2 - 178 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 10.126 ms | 2 - 5 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® SA8775P | 10.907 ms | 2 - 183 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® SA8650P | 10.907 ms | 2 - 183 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® SA8255P | 10.907 ms | 2 - 183 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® QCS8450 | 11.995 ms | 2 - 208 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 10.634 ms | 2 - 12 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 97.214 ms | 13 - 186 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 39.104 ms | 15 - 73 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 5.908 ms | 1 - 175 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® SA7255P | 18.273 ms | 2 - 178 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Qualcomm® SA8295P | 12.854 ms | 2 - 184 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 5.908 ms | 1 - 175 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 4.545 ms | 2 - 184 MB | NPU |
| Segformer-Base | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 39.104 ms | 15 - 73 MB | NPU |
License
- The license for the original implementation of Segformer-Base can be found here.
References
- SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers
- Source Model Implementation
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.
