FFNet-54S: Optimized for Qualcomm Devices

FFNet-54S 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-54S 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 1.24.1	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a8	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.17.0	Download
TFLITE	w8a8	Universal	QAIRT 2.43, TFLite 2.17.0	Download

For more device-specific assets and performance metrics, visit FFNet-54S 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-54S on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Model checkpoint: ffnet54S_dBBB_cityscapes_state_dict_quarts
• Input resolution: 2048x1024
• Number of output classes: 19
• Number of parameters: 18.0M
• Model size (float): 68.8 MB
• Model size (w8a8): 17.5 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
FFNet-54S	ONNX	float	Snapdragon® X Elite	33.964 ms	24 - 24 MB	NPU
FFNet-54S	ONNX	float	Snapdragon® 8 Gen 3 Mobile	24.47 ms	2 - 286 MB	NPU
FFNet-54S	ONNX	float	Qualcomm® QCS8550 (Proxy)	34.774 ms	24 - 27 MB	NPU
FFNet-54S	ONNX	float	Qualcomm® QCS9075	52.658 ms	24 - 51 MB	NPU
FFNet-54S	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	18.096 ms	5 - 204 MB	NPU
FFNet-54S	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	13.506 ms	30 - 256 MB	NPU
FFNet-54S	ONNX	float	Snapdragon® X2 Elite	14.907 ms	22 - 22 MB	NPU
FFNet-54S	ONNX	w8a8	Snapdragon® X Elite	11.185 ms	12 - 12 MB	NPU
FFNet-54S	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	6.937 ms	7 - 266 MB	NPU
FFNet-54S	ONNX	w8a8	Qualcomm® QCS6490	408.555 ms	182 - 237 MB	CPU
FFNet-54S	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	10.557 ms	6 - 9 MB	NPU
FFNet-54S	ONNX	w8a8	Qualcomm® QCS9075	12.906 ms	6 - 9 MB	NPU
FFNet-54S	ONNX	w8a8	Qualcomm® QCM6690	430.965 ms	187 - 196 MB	CPU
FFNet-54S	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	11.638 ms	1 - 196 MB	NPU
FFNet-54S	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	430.383 ms	199 - 209 MB	CPU
FFNet-54S	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	6.931 ms	2 - 205 MB	NPU
FFNet-54S	ONNX	w8a8	Snapdragon® X2 Elite	7.336 ms	13 - 13 MB	NPU
FFNet-54S	QNN_DLC	float	Snapdragon® X Elite	39.581 ms	24 - 24 MB	NPU
FFNet-54S	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	26.788 ms	6 - 295 MB	NPU
FFNet-54S	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	154.346 ms	24 - 226 MB	NPU
FFNet-54S	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	38.469 ms	24 - 26 MB	NPU
FFNet-54S	QNN_DLC	float	Qualcomm® SA8775P	248.409 ms	24 - 224 MB	NPU
FFNet-54S	QNN_DLC	float	Qualcomm® QCS9075	66.894 ms	24 - 52 MB	NPU
FFNet-54S	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	76.861 ms	24 - 306 MB	NPU
FFNet-54S	QNN_DLC	float	Qualcomm® SA7255P	154.346 ms	24 - 226 MB	NPU
FFNet-54S	QNN_DLC	float	Qualcomm® SA8295P	58.63 ms	24 - 221 MB	NPU
FFNet-54S	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	19.516 ms	21 - 243 MB	NPU
FFNet-54S	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	13.656 ms	5 - 247 MB	NPU
FFNet-54S	QNN_DLC	float	Snapdragon® X2 Elite	15.866 ms	24 - 24 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Snapdragon® X Elite	16.749 ms	6 - 6 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	11.172 ms	6 - 265 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® QCS6490	72.051 ms	6 - 14 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	35.439 ms	6 - 205 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	15.82 ms	6 - 7 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® SA8775P	16.281 ms	6 - 206 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® QCS9075	18.81 ms	6 - 14 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® QCM6690	146.247 ms	6 - 242 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	20.081 ms	6 - 264 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® SA7255P	35.439 ms	6 - 205 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Qualcomm® SA8295P	21.42 ms	6 - 208 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	7.588 ms	6 - 220 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	19.882 ms	6 - 217 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	5.479 ms	6 - 243 MB	NPU
FFNet-54S	QNN_DLC	w8a8	Snapdragon® X2 Elite	6.511 ms	6 - 6 MB	NPU
FFNet-54S	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	26.715 ms	2 - 347 MB	NPU
FFNet-54S	TFLITE	float	Qualcomm® QCS8275 (Proxy)	154.571 ms	3 - 228 MB	NPU
FFNet-54S	TFLITE	float	Qualcomm® QCS8550 (Proxy)	38.412 ms	2 - 5 MB	NPU
FFNet-54S	TFLITE	float	Qualcomm® SA8775P	53.647 ms	0 - 223 MB	NPU
FFNet-54S	TFLITE	float	Qualcomm® QCS9075	67.029 ms	0 - 64 MB	NPU
FFNet-54S	TFLITE	float	Qualcomm® QCS8450 (Proxy)	78.369 ms	2 - 335 MB	NPU
FFNet-54S	TFLITE	float	Qualcomm® SA7255P	154.571 ms	3 - 228 MB	NPU
FFNet-54S	TFLITE	float	Qualcomm® SA8295P	58.537 ms	2 - 224 MB	NPU
FFNet-54S	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	19.368 ms	1 - 247 MB	NPU
FFNet-54S	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	13.598 ms	2 - 265 MB	NPU
FFNet-54S	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	5.879 ms	1 - 261 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® QCS6490	55.667 ms	0 - 27 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	22.865 ms	1 - 199 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	8.236 ms	1 - 3 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® SA8775P	8.845 ms	1 - 200 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® QCS9075	10.034 ms	0 - 26 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® QCM6690	118.438 ms	1 - 238 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	13.181 ms	1 - 259 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® SA7255P	22.865 ms	1 - 199 MB	NPU
FFNet-54S	TFLITE	w8a8	Qualcomm® SA8295P	13.128 ms	1 - 202 MB	NPU
FFNet-54S	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	4.441 ms	1 - 217 MB	NPU
FFNet-54S	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	12.701 ms	1 - 218 MB	NPU
FFNet-54S	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	3.195 ms	1 - 237 MB	NPU

License

• The license for the original implementation of FFNet-54S can be found here.

References

• Simple and Efficient Architectures for Semantic Segmentation
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.