Hugging Face's logo

qualcomm
/
FFNet-78S-LowRes

Image Segmentation
PyTorch
real_time
android
Model card Files
xet
 Community

FFNet-78S-LowRes: Optimized for Qualcomm Devices

FFNet-78S-LowRes 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-LowRes 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 1.25.0 Download
ONNX w8a8 Universal QAIRT 2.45, 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-LowRes 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-LowRes on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

  • Model checkpoint: ffnet78S_BCC_cityscapes_state_dict_quarts_pre_down
  • Input resolution: 1024x512
  • Number of output classes: 19
  • Number of parameters: 26.8M
  • Model size (float): 102 MB
  • Model size (w8a8): 26.0 MB

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
FFNet-78S-LowRes ONNX float Snapdragon® X2 Elite 10.07 ms 207 - 207 MB NPU
FFNet-78S-LowRes ONNX float Snapdragon® X Elite 20.045 ms 153 - 153 MB NPU
FFNet-78S-LowRes ONNX float Snapdragon® 8 Gen 3 Mobile 15.429 ms 13 - 253 MB NPU
FFNet-78S-LowRes ONNX float Snapdragon® 8 Gen 1 Mobile 36.007 ms 9 - 228 MB NPU
FFNet-78S-LowRes ONNX float Qualcomm® QCS8550 (Proxy) 20.055 ms 0 - 56 MB NPU
FFNet-78S-LowRes ONNX float Qualcomm® QCS8450 36.007 ms 9 - 228 MB NPU
FFNet-78S-LowRes ONNX float Snapdragon® 8 Elite Mobile 13.252 ms 7 - 190 MB NPU
FFNet-78S-LowRes ONNX float Snapdragon® 8 Elite Gen 5 Mobile 8.035 ms 5 - 188 MB NPU
FFNet-78S-LowRes ONNX float Qualcomm® QCS9075 37.682 ms 6 - 51 MB NPU
FFNet-78S-LowRes ONNX float Qualcomm® QCS8750 13.252 ms 7 - 190 MB NPU
FFNet-78S-LowRes ONNX float Qualcomm® QCS7181 20.045 ms 153 - 153 MB NPU
FFNet-78S-LowRes ONNX w8a8 Snapdragon® X2 Elite 2.717 ms 211 - 211 MB NPU
FFNet-78S-LowRes ONNX w8a8 Snapdragon® X Elite 5.875 ms 148 - 148 MB NPU
FFNet-78S-LowRes ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 3.958 ms 0 - 241 MB NPU
FFNet-78S-LowRes ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 8.202 ms 3 - 245 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCS6490 25.218 ms 2 - 47 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCS8550 (Proxy) 5.656 ms 0 - 33 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCS8450 8.202 ms 3 - 245 MB NPU
FFNet-78S-LowRes ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 2.27 ms 0 - 201 MB NPU
FFNet-78S-LowRes ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 12.315 ms 3 - 212 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCM6690 106.361 ms 0 - 256 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCS9075 5.877 ms 2 - 47 MB NPU
FFNet-78S-LowRes ONNX w8a8 Snapdragon® 8 Elite Mobile 3.046 ms 0 - 197 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCS7790 12.315 ms 3 - 212 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCS8750 3.046 ms 0 - 197 MB NPU
FFNet-78S-LowRes ONNX w8a8 Qualcomm® QCS7181 5.875 ms 148 - 148 MB NPU
FFNet-78S-LowRes QNN_DLC float Snapdragon® X2 Elite 11.048 ms 6 - 6 MB NPU
FFNet-78S-LowRes QNN_DLC float Snapdragon® X Elite 22.721 ms 6 - 6 MB NPU
FFNet-78S-LowRes QNN_DLC float Snapdragon® 8 Gen 3 Mobile 15.747 ms 0 - 239 MB NPU
FFNet-78S-LowRes QNN_DLC float Snapdragon® 8 Gen 1 Mobile 42.025 ms 6 - 234 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® QCS8275 131.764 ms 2 - 183 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® QCS8550 (Proxy) 21.841 ms 6 - 8 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® QCS8450 42.025 ms 6 - 234 MB NPU
FFNet-78S-LowRes QNN_DLC float Snapdragon® 8 Elite Mobile 14.337 ms 6 - 197 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® SA8295P 38.162 ms 0 - 178 MB NPU
FFNet-78S-LowRes QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 8.399 ms 6 - 202 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® SA7255P 131.764 ms 2 - 183 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® QCS9075 38.613 ms 8 - 18 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® QCS8750 14.337 ms 6 - 197 MB NPU
FFNet-78S-LowRes QNN_DLC float Qualcomm® QCS7181 22.721 ms 6 - 6 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Snapdragon® X2 Elite 3.432 ms 2 - 2 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Snapdragon® X Elite 7.643 ms 2 - 2 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 4.879 ms 2 - 129 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 8.546 ms 2 - 130 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS6490 26.387 ms 2 - 5 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS8275 20.503 ms 2 - 80 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS8550 (Proxy) 7.027 ms 0 - 15 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS8450 8.546 ms 2 - 130 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 2.649 ms 2 - 197 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 13.529 ms 2 - 216 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCM6690 106.558 ms 2 - 204 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS9075 7.582 ms 3 - 7 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® SA7255P 20.503 ms 2 - 80 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Snapdragon® 8 Elite Mobile 3.538 ms 2 - 190 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® SA8295P 11.603 ms 1 - 77 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS7790 13.529 ms 2 - 216 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS8750 3.538 ms 2 - 190 MB NPU
FFNet-78S-LowRes QNN_DLC w8a8 Qualcomm® QCS7181 7.643 ms 2 - 2 MB NPU
FFNet-78S-LowRes TFLITE float Snapdragon® 8 Gen 3 Mobile 15.704 ms 0 - 318 MB NPU
FFNet-78S-LowRes TFLITE float Snapdragon® 8 Gen 1 Mobile 41.962 ms 2 - 311 MB NPU
FFNet-78S-LowRes TFLITE float Qualcomm® QCS8275 131.867 ms 3 - 217 MB NPU
FFNet-78S-LowRes TFLITE float Qualcomm® QCS8550 (Proxy) 21.941 ms 2 - 5 MB NPU
FFNet-78S-LowRes TFLITE float Qualcomm® SA8775P 164.885 ms 3 - 32 MB GPU
FFNet-78S-LowRes TFLITE float Qualcomm® SA8650P 164.885 ms 3 - 32 MB GPU
FFNet-78S-LowRes TFLITE float Qualcomm® SA8255P 164.885 ms 3 - 32 MB GPU
FFNet-78S-LowRes TFLITE float Qualcomm® QCS8450 41.962 ms 2 - 311 MB NPU
FFNet-78S-LowRes TFLITE float Snapdragon® 8 Elite Mobile 14.553 ms 1 - 226 MB NPU
FFNet-78S-LowRes TFLITE float Qualcomm® SA8295P 38.202 ms 2 - 212 MB NPU
FFNet-78S-LowRes TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 8.42 ms 0 - 227 MB NPU
FFNet-78S-LowRes TFLITE float Qualcomm® SA7255P 131.867 ms 3 - 217 MB NPU
FFNet-78S-LowRes TFLITE float Qualcomm® QCS9075 38.591 ms 0 - 62 MB NPU
FFNet-78S-LowRes TFLITE float Qualcomm® QCS8750 14.553 ms 1 - 226 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 3.621 ms 1 - 133 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 6.773 ms 0 - 133 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCS6490 22.513 ms 1 - 31 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCS8275 17.352 ms 1 - 78 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCS8550 (Proxy) 4.976 ms 1 - 3 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® SA8775P 167.786 ms 9 - 39 MB GPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® SA8650P 167.786 ms 9 - 39 MB GPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® SA8255P 167.786 ms 9 - 39 MB GPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCS8450 6.773 ms 0 - 133 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 2.081 ms 0 - 194 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 11.708 ms 1 - 211 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCM6690 99.89 ms 1 - 202 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCS9075 5.487 ms 1 - 31 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® SA7255P 17.352 ms 1 - 78 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Snapdragon® 8 Elite Mobile 2.802 ms 0 - 189 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® SA8295P 9.566 ms 1 - 74 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCS7790 11.708 ms 1 - 211 MB NPU
FFNet-78S-LowRes TFLITE w8a8 Qualcomm® QCS8750 2.802 ms 0 - 189 MB NPU

License

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