---
library_name: pytorch
license: other
tags:
- real_time
- android
pipeline_tag: image-segmentation

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_78s_lowres/web-assets/model_demo.png)

# 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](https://github.com/Qualcomm-AI-research/FFNet).
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/ffnet_78s_lowres) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).

Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) 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 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_78s_lowres/releases/v0.51.0/ffnet_78s_lowres-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.42, ONNX Runtime 1.24.3 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_78s_lowres/releases/v0.51.0/ffnet_78s_lowres-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_78s_lowres/releases/v0.51.0/ffnet_78s_lowres-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_78s_lowres/releases/v0.51.0/ffnet_78s_lowres-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_78s_lowres/releases/v0.51.0/ffnet_78s_lowres-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_78s_lowres/releases/v0.51.0/ffnet_78s_lowres-tflite-w8a8.zip)

For more device-specific assets and performance metrics, visit **[FFNet-78S-LowRes on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/ffnet_78s_lowres)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/ffnet_78s_lowres) 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](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/ffnet_78s_lowres) 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® 8 Elite Gen 5 Mobile | 3.821 ms | 7 - 182 MB | NPU
| FFNet-78S-LowRes | ONNX | float | Snapdragon® X2 Elite | 4.22 ms | 47 - 47 MB | NPU
| FFNet-78S-LowRes | ONNX | float | Snapdragon® X Elite | 8.201 ms | 46 - 46 MB | NPU
| FFNet-78S-LowRes | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 5.561 ms | 1 - 208 MB | NPU
| FFNet-78S-LowRes | ONNX | float | Qualcomm® QCS8550 (Proxy) | 7.809 ms | 6 - 9 MB | NPU
| FFNet-78S-LowRes | ONNX | float | Qualcomm® QCS9075 | 12.879 ms | 6 - 15 MB | NPU
| FFNet-78S-LowRes | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.566 ms | 2 - 176 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 1.273 ms | 0 - 61 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Snapdragon® X2 Elite | 1.323 ms | 25 - 25 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Snapdragon® X Elite | 2.995 ms | 25 - 25 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.989 ms | 0 - 119 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Qualcomm® QCS6490 | 109.088 ms | 56 - 117 MB | CPU
| FFNet-78S-LowRes | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 2.835 ms | 0 - 57 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Qualcomm® QCS9075 | 3.328 ms | 1 - 4 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Qualcomm® QCM6690 | 118.401 ms | 60 - 69 MB | CPU
| FFNet-78S-LowRes | ONNX | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.506 ms | 0 - 62 MB | NPU
| FFNet-78S-LowRes | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 119.078 ms | 70 - 81 MB | CPU
| FFNet-78S-LowRes | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.57 ms | 6 - 55 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Snapdragon® X2 Elite | 7.612 ms | 6 - 6 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Snapdragon® X Elite | 15.465 ms | 6 - 6 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 9.789 ms | 6 - 80 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 51.219 ms | 1 - 43 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 14.74 ms | 6 - 204 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Qualcomm® SA8775P | 19.184 ms | 1 - 45 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Qualcomm® QCS9075 | 18.269 ms | 6 - 14 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 26.575 ms | 6 - 72 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Qualcomm® SA7255P | 51.219 ms | 1 - 43 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Qualcomm® SA8295P | 20.95 ms | 0 - 35 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 8.075 ms | 6 - 52 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 1.901 ms | 2 - 58 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 2.191 ms | 2 - 2 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Snapdragon® X Elite | 5.116 ms | 2 - 2 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 3.352 ms | 2 - 104 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 15.029 ms | 4 - 7 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® QCS8275 (Proxy) | 10.655 ms | 2 - 56 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 4.774 ms | 2 - 24 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® SA8775P | 5.236 ms | 2 - 59 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 7.009 ms | 3 - 6 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 35.608 ms | 2 - 193 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® QCS8450 (Proxy) | 7.16 ms | 2 - 100 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® SA7255P | 10.655 ms | 2 - 56 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Qualcomm® SA8295P | 6.676 ms | 1 - 54 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 2.351 ms | 2 - 58 MB | NPU
| FFNet-78S-LowRes | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 7.333 ms | 2 - 188 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 6.578 ms | 1 - 79 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 9.852 ms | 1 - 155 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 51.052 ms | 1 - 76 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 14.558 ms | 1 - 4 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Qualcomm® SA8775P | 19.114 ms | 1 - 77 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Qualcomm® QCS9075 | 17.628 ms | 0 - 60 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 26.242 ms | 0 - 146 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Qualcomm® SA7255P | 51.052 ms | 1 - 76 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Qualcomm® SA8295P | 21.02 ms | 1 - 67 MB | NPU
| FFNet-78S-LowRes | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 8.123 ms | 0 - 82 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 1.334 ms | 0 - 54 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 2.049 ms | 0 - 105 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® QCS6490 | 9.281 ms | 0 - 30 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® QCS8275 (Proxy) | 7.463 ms | 0 - 51 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 2.865 ms | 0 - 4 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® SA8775P | 3.365 ms | 0 - 54 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® QCS9075 | 3.369 ms | 0 - 29 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® QCM6690 | 30.986 ms | 0 - 189 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® QCS8450 (Proxy) | 3.617 ms | 0 - 100 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® SA7255P | 7.463 ms | 0 - 51 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Qualcomm® SA8295P | 4.543 ms | 0 - 49 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Snapdragon® 8 Elite For Galaxy Mobile | 1.555 ms | 0 - 54 MB | NPU
| FFNet-78S-LowRes | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 4.452 ms | 0 - 180 MB | NPU

## License
* The license for the original implementation of FFNet-78S-LowRes can be found
  [here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE).

## References
* [Simple and Efficient Architectures for Semantic Segmentation](https://arxiv.org/abs/2206.08236)
* [Source Model Implementation](https://github.com/Qualcomm-AI-research/FFNet)

## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).