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 2.42, ONNX Runtime 1.25.0 Download
ONNX w8a8 Universal QAIRT 2.42, ONNX Runtime 1.25.0 Download
QNN_DLC float 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 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:

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

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 Mobile 17.942 ms 29 - 258 MB NPU
FFNet-78S ONNX float Snapdragon® X2 Elite 18.143 ms 188 - 188 MB NPU
FFNet-78S ONNX float Snapdragon® X Elite 38.156 ms 157 - 157 MB NPU
FFNet-78S ONNX float Snapdragon® 8 Gen 3 Mobile 27.076 ms 29 - 332 MB NPU
FFNet-78S ONNX float Qualcomm® QCS8550 (Proxy) 38.132 ms 4 - 80 MB NPU
FFNet-78S ONNX float Snapdragon® 8 Elite For Galaxy Mobile 19.935 ms 6 - 215 MB NPU
FFNet-78S ONNX float Qualcomm® QCS9075 59.514 ms 24 - 93 MB NPU
FFNet-78S ONNX float Qualcomm® QCS8750 19.935 ms 6 - 215 MB NPU
FFNet-78S ONNX float Qualcomm® QCS7181 38.156 ms 157 - 157 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 7.294 ms 6 - 227 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® X2 Elite 7.881 ms 206 - 206 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® X Elite 14.751 ms 175 - 175 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 11.686 ms 7 - 299 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® QCS6490 493.914 ms 162 - 227 MB CPU
FFNet-78S ONNX w8a8 Qualcomm® QCS8550 (Proxy) 15.498 ms 0 - 32 MB NPU
FFNet-78S ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 534.674 ms 147 - 158 MB CPU
FFNet-78S ONNX w8a8 Qualcomm® QCM6690 534.392 ms 132 - 142 MB CPU
FFNet-78S ONNX w8a8 Snapdragon® 8 Elite For Galaxy Mobile 8.785 ms 1 - 213 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® QCS9075 14.523 ms 6 - 51 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® QCS7790 534.674 ms 147 - 158 MB CPU
FFNet-78S ONNX w8a8 Qualcomm® QCS8750 8.785 ms 1 - 213 MB NPU
FFNet-78S ONNX w8a8 Qualcomm® QCS7181 14.751 ms 175 - 175 MB NPU
FFNet-78S QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 15.487 ms 24 - 274 MB NPU
FFNet-78S QNN_DLC float Snapdragon® X2 Elite 18.085 ms 24 - 24 MB NPU
FFNet-78S QNN_DLC float Snapdragon® X Elite 44.623 ms 24 - 24 MB NPU
FFNet-78S QNN_DLC float Snapdragon® 8 Gen 3 Mobile 29.286 ms 23 - 319 MB NPU
FFNet-78S QNN_DLC float Qualcomm® QCS8275 187.964 ms 24 - 234 MB NPU
FFNet-78S QNN_DLC float Qualcomm® QCS8550 (Proxy) 42.569 ms 24 - 26 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8775P 60.877 ms 24 - 234 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8650P 60.877 ms 24 - 234 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8255P 60.877 ms 24 - 234 MB NPU
FFNet-78S QNN_DLC float Qualcomm® QCS8450 (Proxy) 86.157 ms 3 - 284 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA8295P 66.503 ms 24 - 229 MB NPU
FFNet-78S QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 21.849 ms 10 - 237 MB NPU
FFNet-78S QNN_DLC float Qualcomm® SA7255P 187.964 ms 24 - 234 MB NPU
FFNet-78S QNN_DLC float Qualcomm® QCS9075 73.415 ms 24 - 52 MB NPU
FFNet-78S QNN_DLC float Qualcomm® QCS8750 21.849 ms 10 - 237 MB NPU
FFNet-78S QNN_DLC float Qualcomm® QCS7181 44.623 ms 24 - 24 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 5.877 ms 6 - 249 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® X2 Elite 6.996 ms 6 - 6 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® X Elite 17.8 ms 6 - 6 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 11.601 ms 6 - 293 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS6490 72.639 ms 6 - 14 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS8275 39.064 ms 6 - 225 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS8550 (Proxy) 16.717 ms 6 - 183 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8775P 17.178 ms 6 - 226 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8650P 17.178 ms 6 - 226 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8255P 17.178 ms 6 - 226 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA7255P 39.064 ms 6 - 225 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 22.422 ms 6 - 242 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® SA8295P 23.335 ms 6 - 227 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCM6690 163.688 ms 6 - 254 MB NPU
FFNet-78S QNN_DLC w8a8 Snapdragon® 8 Elite For Galaxy Mobile 8.053 ms 6 - 239 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS9075 19.731 ms 6 - 14 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS8450 (Proxy) 23.828 ms 6 - 291 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS7790 22.422 ms 6 - 242 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS8750 8.053 ms 6 - 239 MB NPU
FFNet-78S QNN_DLC w8a8 Qualcomm® QCS7181 17.8 ms 6 - 6 MB NPU
FFNet-78S TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 15.437 ms 2 - 285 MB NPU
FFNet-78S TFLITE float Snapdragon® 8 Gen 3 Mobile 29.522 ms 2 - 376 MB NPU
FFNet-78S TFLITE float Qualcomm® QCS8275 188.04 ms 3 - 245 MB NPU
FFNet-78S TFLITE float Qualcomm® QCS8550 (Proxy) 42.824 ms 3 - 5 MB NPU
FFNet-78S TFLITE float Qualcomm® SA8775P 60.995 ms 3 - 245 MB NPU
FFNet-78S TFLITE float Qualcomm® SA8650P 60.995 ms 3 - 245 MB NPU
FFNet-78S TFLITE float Qualcomm® SA8255P 60.995 ms 3 - 245 MB NPU
FFNet-78S TFLITE float Qualcomm® QCS8450 (Proxy) 86.527 ms 2 - 361 MB NPU
FFNet-78S TFLITE float Qualcomm® SA8295P 66.479 ms 3 - 240 MB NPU
FFNet-78S TFLITE float Snapdragon® 8 Elite For Galaxy Mobile 21.839 ms 1 - 261 MB NPU
FFNet-78S TFLITE float Qualcomm® SA7255P 188.04 ms 3 - 245 MB NPU
FFNet-78S TFLITE float Qualcomm® QCS9075 73.752 ms 0 - 82 MB NPU
FFNet-78S TFLITE float Qualcomm® QCS8750 21.839 ms 1 - 261 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 3.576 ms 0 - 240 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 6.439 ms 1 - 287 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS6490 57.168 ms 1 - 36 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS8275 26.224 ms 1 - 218 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS8550 (Proxy) 9.072 ms 1 - 3 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8775P 9.743 ms 1 - 219 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8650P 9.743 ms 1 - 219 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8255P 9.743 ms 1 - 219 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA7255P 26.224 ms 1 - 218 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 14.576 ms 1 - 234 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® SA8295P 14.889 ms 1 - 222 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCM6690 141.187 ms 1 - 246 MB NPU
FFNet-78S TFLITE w8a8 Snapdragon® 8 Elite For Galaxy Mobile 5.046 ms 1 - 233 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS9075 11.171 ms 0 - 36 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS8450 (Proxy) 18.378 ms 1 - 289 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS7790 14.576 ms 1 - 234 MB NPU
FFNet-78S TFLITE w8a8 Qualcomm® QCS8750 5.046 ms 1 - 233 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