Hugging Face's logo

qualcomm
/
FFNet-78S

Image Segmentation
PyTorch
android
Model card Files
xet
 Community

FFNet-78S: Optimized for Qualcomm Devices

FFNet-78S 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-78S 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 1.30.0 Download
ONNX w8a8 Universal QAIRT 2.50, ONNX Runtime 1.30.0 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-78S 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-78S on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

  • Input resolution: 2048x1024
  • Model checkpoint: ffnet78S_dBBB_cityscapes_state_dict_quarts
  • Model size (float): 105 MB
  • Model size (w8a8): 26.7 MB
  • Number of output classes: 19
  • Number of parameters: 27.5M

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
FFNet-78S ONNX float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 16.22 ms 5 - 220 MB NPU
FFNet-78S ONNX float Snapdragon® 8 Elite For Galaxy Mobile 20.831 ms 6 - 196 MB NPU
FFNet-78S ONNX float Snapdragon® X2 Elite 17.501 ms 24 - 24 MB NPU
FFNet-78S ONNX float Snapdragon® X Elite 36.262 ms 31 - 31 MB NPU
FFNet-78S ONNX float Snapdragon® 8 Gen 3 Mobile 25.923 ms 13 - 296 MB NPU
FFNet-78S ONNX float Snapdragon® 8 Gen 1 Mobile 59.782 ms 0 - 272 MB NPU
FFNet-78S ONNX float Qualcomm® Dragonwing™ IQ-8275 57.611 ms 24 - 52 MB NPU
FFNet-78S ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 37.14 ms 24 - 54 MB NPU
FFNet-78S ONNX float Qualcomm® QCS8450 59.782 ms 0 - 272 MB NPU
FFNet-78S ONNX float Qualcomm® Dragonwing™ IQ-9075 55.024 ms 24 - 51 MB NPU
FFNet-78S ONNX float Qualcomm® Dragonwing™ IQ-X7181 36.262 ms 31 - 31 MB NPU
FFNet-78S ONNX float Qualcomm® Dragonwing™ Q-8750 20.831 ms 6 - 196 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4.247 ms 2 - 210 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 5.731 ms 1 - 208 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® X2 Elite 4.729 ms 7 - 7 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® X Elite 12.025 ms 22 - 22 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 7.585 ms 9 - 291 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 16.285 ms 7 - 291 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ QCS6490 49.994 ms 6 - 9 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ IQ-8275 11.336 ms 6 - 10 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 11.385 ms 6 - 10 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® QCS8450 16.285 ms 7 - 291 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ IQ-9075 12.767 ms 6 - 9 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ IQ-X7181 12.025 ms 22 - 22 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ Q-6690 142.371 ms 5 - 230 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ Q-7790 16.38 ms 7 - 237 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® Dragonwing™ Q-8750 5.731 ms 1 - 208 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 16.38 ms 7 - 237 MB NPU
FFNet-78S QNN_DLC float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 15.296 ms 3 - 231 MB NPU
FFNet-78S QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 21.931 ms 7 - 223 MB NPU
FFNet-78S QNN_DLC float Snapdragon® X2 Elite 17.992 ms 24 - 24 MB NPU
FFNet-78S QNN_DLC float Snapdragon® X Elite 43.606 ms 24 - 24 MB NPU
FFNet-78S QNN_DLC float Snapdragon® 8 Gen 3 Mobile 29.431 ms 16 - 298 MB NPU
FFNet-78S QNN_DLC float Snapdragon® 8 Gen 1 Mobile 60.678 ms 25 - 298 MB NPU
FFNet-78S QNN_DLC float Qualcomm® Dragonwing™ IQ-8275 62.415 ms 24 - 53 MB NPU
FFNet-78S QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 42.215 ms 24 - 355 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8775P 60.693 ms 24 - 221 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8650P 60.693 ms 24 - 221 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8255P 60.693 ms 24 - 221 MB NPU
FFNet-78S QNN_DLC float Qualcomm® QCS8450 60.678 ms 25 - 298 MB NPU
FFNet-78S QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 61.509 ms 24 - 53 MB NPU
FFNet-78S QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 43.606 ms 24 - 24 MB NPU
FFNet-78S QNN_DLC float Qualcomm® Dragonwing™ Q-8750 21.931 ms 7 - 223 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8295P 66.246 ms 24 - 217 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 6.062 ms 6 - 240 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 8.169 ms 6 - 226 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® X2 Elite 7.03 ms 6 - 6 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® X Elite 17.624 ms 6 - 6 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 11.724 ms 6 - 282 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 19.736 ms 6 - 282 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS6490 53.223 ms 6 - 15 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-8275 16.579 ms 3 - 12 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 16.615 ms 6 - 8 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8775P 17.26 ms 6 - 215 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8650P 17.26 ms 6 - 215 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8255P 17.26 ms 6 - 215 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS8450 19.736 ms 6 - 282 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-9075 17.688 ms 6 - 15 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-X7181 17.624 ms 6 - 6 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-6690 143.719 ms 6 - 220 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-7790 21.891 ms 6 - 233 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-8750 8.169 ms 6 - 226 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA7255P 38.988 ms 6 - 214 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8295P 23.28 ms 6 - 218 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 21.891 ms 6 - 233 MB NPU
FFNet-78S TFLITE float Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 15.35 ms 2 - 260 MB NPU
FFNet-78S TFLITE float Snapdragon® 8 Elite For Galaxy Mobile 21.82 ms 2 - 245 MB NPU
FFNet-78S TFLITE float Snapdragon® 8 Gen 3 Mobile 29.647 ms 1 - 363 MB NPU
FFNet-78S TFLITE float Snapdragon® 8 Gen 1 Mobile 60.842 ms 3 - 353 MB NPU
FFNet-78S TFLITE float Qualcomm® Dragonwing™ IQ-8275 62.831 ms 0 - 83 MB NPU
FFNet-78S TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 42.429 ms 2 - 5 MB NPU
FFNet-78S TFLITE float Qualcomm® QCS8450 60.842 ms 3 - 353 MB NPU
FFNet-78S TFLITE float Qualcomm® Dragonwing™ IQ-9075 61.939 ms 0 - 83 MB NPU
FFNet-78S TFLITE float Qualcomm® Dragonwing™ Q-8750 21.82 ms 2 - 245 MB NPU
FFNet-78S TFLITE float Qualcomm® SA7255P 188.683 ms 3 - 230 MB NPU
FFNet-78S TFLITE float Qualcomm® SA8295P 66.254 ms 3 - 229 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 3.745 ms 0 - 232 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 4.947 ms 1 - 219 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 6.471 ms 1 - 279 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 11.395 ms 1 - 279 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® Dragonwing™ QCS6490 38.238 ms 1 - 37 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® Dragonwing™ IQ-8275 9.759 ms 1 - 38 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 9.021 ms 1 - 3 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8775P 9.761 ms 1 - 209 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8650P 9.761 ms 1 - 209 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8255P 9.761 ms 1 - 209 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS8450 11.395 ms 1 - 279 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® Dragonwing™ IQ-9075 9.581 ms 1 - 37 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® Dragonwing™ Q-6690 147.864 ms 1 - 212 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® Dragonwing™ Q-7790 14.408 ms 1 - 224 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® Dragonwing™ Q-8750 4.947 ms 1 - 219 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA7255P 26.206 ms 1 - 208 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8295P 14.886 ms 1 - 212 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 14.408 ms 1 - 224 MB NPU

License

  • The license for the original implementation of FFNet-78S 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-78S