Hugging Face's logo

qualcomm
/
FFNet-40S

Image Segmentation
PyTorch
real_time
android
Model card Files
xet
 Community

FFNet-40S: Optimized for Qualcomm Devices

FFNet-40S is a "fuss-free network" that segments street scene images with per-pixel classes like road, sidewalk, and pedestrian. Trained on the Cityscapes dataset.

This is based on the implementation of FFNet-40S 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 2.50, ONNX Runtime 1.27.1 Download
ONNX w8a8 Universal QAIRT 2.50, ONNX Runtime 1.27.1 Download
QNN_DLC float Universal QAIRT 2.50 Download
QNN_DLC w8a8 Universal QAIRT 2.50 Download
TFLITE float Universal QAIRT 2.50 Download
TFLITE w8a8 Universal QAIRT 2.50 Download

For more device-specific assets and performance metrics, visit FFNet-40S 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 FFNet-40S on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

  • Input resolution: 2048x1024
  • Model checkpoint: ffnet40S_dBBB_cityscapes_state_dict_quarts
  • Model size (float): 53.1 MB
  • Model size (w8a8): 13.5 MB
  • Number of output classes: 19
  • Number of parameters: 13.9M

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
FFNet-40S ONNX float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 13.141 ms 5 - 205 MB NPU
FFNet-40S ONNX float Snapdragon® 8 Elite For Galaxy Mobile 16.96 ms 6 - 184 MB NPU
FFNet-40S ONNX float Snapdragon® X2 Elite 13.99 ms 24 - 24 MB NPU
FFNet-40S ONNX float Snapdragon® X Elite 30.337 ms 24 - 24 MB NPU
FFNet-40S ONNX float Snapdragon® 8 Gen 3 Mobile 25.16 ms 31 - 288 MB NPU
FFNet-40S ONNX float Snapdragon® 8 Gen 1 Mobile 46.781 ms 0 - 251 MB NPU
FFNet-40S ONNX float Qualcomm® Dragonwing™ IQ-8275 43.562 ms 24 - 28 MB NPU
FFNet-40S ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 31.418 ms 24 - 340 MB NPU
FFNet-40S ONNX float Qualcomm® QCS8450 46.781 ms 0 - 251 MB NPU
FFNet-40S ONNX float Qualcomm® Dragonwing™ IQ-9075 44.382 ms 24 - 27 MB NPU
FFNet-40S ONNX float Qualcomm® Dragonwing™ IQ-X7181 30.337 ms 24 - 24 MB NPU
FFNet-40S ONNX float Qualcomm® Dragonwing™ Q-8750 16.96 ms 6 - 184 MB NPU
FFNet-40S ONNX w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 3.535 ms 1 - 190 MB NPU
FFNet-40S ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 4.92 ms 2 - 185 MB NPU
FFNet-40S ONNX w8a8 Snapdragon® X2 Elite 3.897 ms 7 - 7 MB NPU
FFNet-40S ONNX w8a8 Snapdragon® X Elite 10.632 ms 8 - 8 MB NPU
FFNet-40S ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 7.142 ms 7 - 244 MB NPU
FFNet-40S ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 13.992 ms 7 - 246 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ QCS6490 43.023 ms 6 - 9 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ IQ-8275 10.439 ms 6 - 10 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 10.001 ms 0 - 11 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® QCS8450 13.992 ms 7 - 246 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ IQ-9075 11.63 ms 6 - 9 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ IQ-X7181 10.632 ms 8 - 8 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ Q-6690 115.809 ms 6 - 200 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ Q-7790 14.667 ms 7 - 204 MB NPU
FFNet-40S ONNX w8a8 Qualcomm® Dragonwing™ Q-8750 4.92 ms 2 - 185 MB NPU
FFNet-40S ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 14.667 ms 7 - 204 MB NPU
FFNet-40S QNN_DLC float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 12.458 ms 24 - 242 MB NPU
FFNet-40S QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 18.195 ms 23 - 226 MB NPU
FFNet-40S QNN_DLC float Snapdragon® X2 Elite 14.714 ms 24 - 24 MB NPU
FFNet-40S QNN_DLC float Snapdragon® X Elite 37.888 ms 24 - 24 MB NPU
FFNet-40S QNN_DLC float Snapdragon® 8 Gen 3 Mobile 25.889 ms 24 - 282 MB NPU
FFNet-40S QNN_DLC float Snapdragon® 8 Gen 1 Mobile 50.29 ms 24 - 279 MB NPU
FFNet-40S QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 48.881 ms 24 - 53 MB NPU
FFNet-40S QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 36.545 ms 24 - 256 MB NPU
FFNet-40S QNN_DLC float Qualcomm® SA8775P 49.213 ms 24 - 210 MB NPU
FFNet-40S QNN_DLC float Qualcomm® SA8650P 49.213 ms 24 - 210 MB NPU
FFNet-40S QNN_DLC float Qualcomm® SA8255P 49.213 ms 24 - 210 MB NPU
FFNet-40S QNN_DLC float Qualcomm® QCS8450 50.29 ms 24 - 279 MB NPU
FFNet-40S QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 50.21 ms 24 - 53 MB NPU
FFNet-40S QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 37.888 ms 24 - 24 MB NPU
FFNet-40S QNN_DLC float Qualcomm® Dragonwing™ Q-8750 18.195 ms 23 - 226 MB NPU
FFNet-40S QNN_DLC float Qualcomm® SA8295P 53.71 ms 24 - 212 MB NPU
FFNet-40S QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 5.382 ms 6 - 222 MB NPU
FFNet-40S QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 7.187 ms 6 - 208 MB NPU
FFNet-40S QNN_DLC w8a8 Snapdragon® X2 Elite 6.286 ms 6 - 6 MB NPU
FFNet-40S QNN_DLC w8a8 Snapdragon® X Elite 16.15 ms 6 - 6 MB NPU
FFNet-40S QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 10.512 ms 6 - 240 MB NPU
FFNet-40S QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 17.874 ms 6 - 243 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS6490 49.245 ms 1 - 10 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-8275 15.011 ms 1 - 11 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 15.184 ms 6 - 8 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® SA8650P 15.675 ms 6 - 203 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® SA8255P 15.675 ms 6 - 203 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® QCS8450 17.874 ms 6 - 243 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-9075 15.809 ms 6 - 15 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-X7181 16.15 ms 6 - 6 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-6690 127.295 ms 6 - 194 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-7790 19.129 ms 6 - 206 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-8750 7.187 ms 6 - 208 MB NPU
FFNet-40S QNN_DLC w8a8 Qualcomm® SA8295P 20.138 ms 6 - 196 MB NPU
FFNet-40S QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 19.129 ms 6 - 206 MB NPU
FFNet-40S TFLITE float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 12.382 ms 2 - 231 MB NPU
FFNet-40S TFLITE float Snapdragon® 8 Elite For Galaxy Mobile 18.086 ms 1 - 216 MB NPU
FFNet-40S TFLITE float Snapdragon® 8 Gen 3 Mobile 26.184 ms 2 - 297 MB NPU
FFNet-40S TFLITE float Snapdragon® 8 Gen 1 Mobile 49.909 ms 2 - 294 MB NPU
FFNet-40S TFLITE float Qualcomm® Dragonwing™ IQ-8275 49.082 ms 0 - 57 MB NPU
FFNet-40S TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 36.26 ms 3 - 278 MB NPU
FFNet-40S TFLITE float Qualcomm® SA8775P 49.221 ms 2 - 202 MB NPU
FFNet-40S TFLITE float Qualcomm® SA8650P 49.221 ms 2 - 202 MB NPU
FFNet-40S TFLITE float Qualcomm® SA8255P 49.221 ms 2 - 202 MB NPU
FFNet-40S TFLITE float Qualcomm® QCS8450 49.909 ms 2 - 294 MB NPU
FFNet-40S TFLITE float Qualcomm® Dragonwing™ IQ-9075 50.144 ms 0 - 56 MB NPU
FFNet-40S TFLITE float Qualcomm® Dragonwing™ Q-8750 18.086 ms 1 - 216 MB NPU
FFNet-40S TFLITE float Qualcomm® SA8295P 53.94 ms 0 - 201 MB NPU
FFNet-40S TFLITE w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 3.06 ms 1 - 214 MB NPU
FFNet-40S TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 4.018 ms 0 - 202 MB NPU
FFNet-40S TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 5.337 ms 0 - 238 MB NPU
FFNet-40S TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 9.417 ms 1 - 239 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® Dragonwing™ QCS6490 33.14 ms 1 - 24 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® Dragonwing™ IQ-8275 8.094 ms 1 - 24 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 7.496 ms 1 - 3 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® SA8650P 8.267 ms 1 - 197 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® SA8255P 8.267 ms 1 - 197 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® QCS8450 9.417 ms 1 - 239 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® Dragonwing™ IQ-9075 8.045 ms 1 - 24 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® Dragonwing™ Q-6690 105.39 ms 0 - 186 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® Dragonwing™ Q-7790 11.359 ms 1 - 200 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® Dragonwing™ Q-8750 4.018 ms 0 - 202 MB NPU
FFNet-40S TFLITE w8a8 Qualcomm® SA8295P 11.725 ms 0 - 190 MB NPU
FFNet-40S TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 11.359 ms 1 - 200 MB NPU

License

  • The license for the original implementation of FFNet-40S can be found here.

References

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Image Segmentation
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for qualcomm/FFNet-40S