FFNet-40S: Optimized for Qualcomm Devices

FFNet-40S 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-40S 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 1.24.3	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a8	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.19.1	Download
TFLITE	w8a8	Universal	QAIRT 2.43, TFLite 2.19.1	Download

For more device-specific assets and performance metrics, visit FFNet-40S 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-40S on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Model checkpoint: ffnet40S_dBBB_cityscapes_state_dict_quarts
• Input resolution: 2048x1024
• Number of output classes: 19
• Number of parameters: 13.9M
• Model size (float): 53.1 MB
• Model size (w8a8): 13.5 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
FFNet-40S	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	12.343 ms	29 - 255 MB	NPU
FFNet-40S	ONNX	float	Snapdragon® X2 Elite	13.463 ms	22 - 22 MB	NPU
FFNet-40S	ONNX	float	Snapdragon® X Elite	31.619 ms	24 - 24 MB	NPU
FFNet-40S	ONNX	float	Snapdragon® 8 Gen 3 Mobile	22.658 ms	31 - 305 MB	NPU
FFNet-40S	ONNX	float	Qualcomm® QCS8550 (Proxy)	31.664 ms	24 - 37 MB	NPU
FFNet-40S	ONNX	float	Qualcomm® QCS9075	47.74 ms	24 - 27 MB	NPU
FFNet-40S	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	15.997 ms	6 - 204 MB	NPU
FFNet-40S	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	6.546 ms	0 - 195 MB	NPU
FFNet-40S	ONNX	w8a8	Snapdragon® X2 Elite	6.975 ms	9 - 9 MB	NPU
FFNet-40S	ONNX	w8a8	Snapdragon® X Elite	10.509 ms	8 - 8 MB	NPU
FFNet-40S	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	6.558 ms	7 - 248 MB	NPU
FFNet-40S	ONNX	w8a8	Qualcomm® QCS6490	364.807 ms	203 - 237 MB	CPU
FFNet-40S	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	9.818 ms	0 - 13 MB	NPU
FFNet-40S	ONNX	w8a8	Qualcomm® QCS9075	13.148 ms	6 - 9 MB	NPU
FFNet-40S	ONNX	w8a8	Qualcomm® QCM6690	365.91 ms	197 - 205 MB	CPU
FFNet-40S	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	7.73 ms	1 - 192 MB	NPU
FFNet-40S	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	361.867 ms	168 - 178 MB	CPU
FFNet-40S	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	12.608 ms	16 - 254 MB	NPU
FFNet-40S	QNN_DLC	float	Snapdragon® X2 Elite	14.559 ms	24 - 24 MB	NPU
FFNet-40S	QNN_DLC	float	Snapdragon® X Elite	37.145 ms	24 - 24 MB	NPU
FFNet-40S	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	24.775 ms	22 - 301 MB	NPU
FFNet-40S	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	134.581 ms	24 - 219 MB	NPU
FFNet-40S	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	36.057 ms	24 - 293 MB	NPU
FFNet-40S	QNN_DLC	float	Qualcomm® SA8775P	48.845 ms	15 - 212 MB	NPU
FFNet-40S	QNN_DLC	float	Qualcomm® QCS9075	61.884 ms	24 - 52 MB	NPU
FFNet-40S	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	113.56 ms	24 - 308 MB	NPU
FFNet-40S	QNN_DLC	float	Qualcomm® SA7255P	134.581 ms	24 - 219 MB	NPU
FFNet-40S	QNN_DLC	float	Qualcomm® SA8295P	64.958 ms	24 - 224 MB	NPU
FFNet-40S	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	18.075 ms	12 - 238 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	5.211 ms	6 - 236 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Snapdragon® X2 Elite	6.2 ms	6 - 6 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Snapdragon® X Elite	15.993 ms	6 - 6 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	10.579 ms	6 - 250 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® QCS6490	67.129 ms	6 - 14 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	32.892 ms	6 - 198 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	15.12 ms	6 - 21 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® SA8775P	15.746 ms	6 - 199 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® QCS9075	18.703 ms	6 - 14 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® QCM6690	127.255 ms	6 - 238 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	21.414 ms	6 - 249 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® SA7255P	32.892 ms	6 - 198 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Qualcomm® SA8295P	20.196 ms	6 - 201 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	7.227 ms	6 - 215 MB	NPU
FFNet-40S	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	18.89 ms	6 - 221 MB	NPU
FFNet-40S	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	12.636 ms	2 - 252 MB	NPU
FFNet-40S	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	24.679 ms	1 - 316 MB	NPU
FFNet-40S	TFLITE	float	Qualcomm® QCS8275 (Proxy)	134.669 ms	3 - 213 MB	NPU
FFNet-40S	TFLITE	float	Qualcomm® QCS8550 (Proxy)	36.061 ms	2 - 4 MB	NPU
FFNet-40S	TFLITE	float	Qualcomm® SA8775P	48.819 ms	2 - 213 MB	NPU
FFNet-40S	TFLITE	float	Qualcomm® QCS9075	61.932 ms	0 - 56 MB	NPU
FFNet-40S	TFLITE	float	Qualcomm® QCS8450 (Proxy)	113.882 ms	3 - 324 MB	NPU
FFNet-40S	TFLITE	float	Qualcomm® SA7255P	134.669 ms	3 - 213 MB	NPU
FFNet-40S	TFLITE	float	Qualcomm® SA8295P	64.901 ms	2 - 222 MB	NPU
FFNet-40S	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	18.072 ms	1 - 242 MB	NPU
FFNet-40S	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	2.945 ms	1 - 229 MB	NPU
FFNet-40S	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	5.443 ms	1 - 244 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® QCS6490	52.979 ms	1 - 23 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	20.422 ms	1 - 191 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	7.612 ms	1 - 4 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® SA8775P	8.269 ms	1 - 193 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® QCS9075	9.558 ms	0 - 22 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® QCM6690	103.066 ms	0 - 230 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	12.21 ms	1 - 246 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® SA7255P	20.422 ms	1 - 191 MB	NPU
FFNet-40S	TFLITE	w8a8	Qualcomm® SA8295P	11.818 ms	1 - 195 MB	NPU
FFNet-40S	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	4.079 ms	1 - 207 MB	NPU
FFNet-40S	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	11.575 ms	1 - 212 MB	NPU

License

• The license for the original implementation of FFNet-40S 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.