PidNet: Optimized for Qualcomm Devices

PIDNet (Proportional-Integral-Derivative Network) is a real-time semantic segmentation model based on PID controllers

This is based on the implementation of PidNet 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 2.37, ONNX Runtime 1.23.0	Download
ONNX	w8a8	Universal	QAIRT 2.37, ONNX Runtime 1.23.0	Download
QNN_DLC	float	Universal	QAIRT 2.42	Download
QNN_DLC	w8a8	Universal	QAIRT 2.42	Download
TFLITE	float	Universal	QAIRT 2.42, TFLite 2.17.0	Download
TFLITE	w8a8	Universal	QAIRT 2.42, TFLite 2.17.0	Download

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

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

Model checkpoint: PIDNet_S_Cityscapes_val.pt
Inference latency: RealTime
Input resolution: 1024x2048
Number of output classes: 19
Number of parameters: 8.06M
Model size (float): 29.1 MB
Model size (w8a8): 8.02 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
PidNet	ONNX	float	Snapdragon® X Elite	29.672 ms	24 - 24 MB	NPU
PidNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	20.062 ms	30 - 283 MB	NPU
PidNet	ONNX	float	Qualcomm® QCS8550 (Proxy)	29.443 ms	24 - 47 MB	NPU
PidNet	ONNX	float	Qualcomm® QCS9075	46.981 ms	24 - 50 MB	NPU
PidNet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	16.219 ms	6 - 170 MB	NPU
PidNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	11.504 ms	10 - 230 MB	NPU
PidNet	ONNX	w8a8	Snapdragon® X Elite	62.888 ms	131 - 131 MB	NPU
PidNet	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	47.162 ms	104 - 316 MB	NPU
PidNet	ONNX	w8a8	Qualcomm® QCS6490	396.442 ms	197 - 217 MB	CPU
PidNet	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	61.97 ms	99 - 102 MB	NPU
PidNet	ONNX	w8a8	Qualcomm® QCS9075	66.445 ms	100 - 102 MB	NPU
PidNet	ONNX	w8a8	Qualcomm® QCM6690	352.536 ms	198 - 207 MB	CPU
PidNet	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	41.36 ms	100 - 258 MB	NPU
PidNet	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	328.507 ms	182 - 192 MB	CPU
PidNet	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	41.2 ms	96 - 263 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® X Elite	40.667 ms	24 - 24 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	26.571 ms	24 - 340 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	120.917 ms	24 - 243 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	39.47 ms	24 - 26 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® SA8775P	48.183 ms	24 - 244 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS9075	61.963 ms	24 - 52 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	77.089 ms	5 - 336 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® SA7255P	120.917 ms	24 - 243 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® SA8295P	53.509 ms	24 - 261 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	19.353 ms	15 - 261 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	12.477 ms	24 - 296 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® X Elite	60.343 ms	6 - 6 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	43.386 ms	6 - 271 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	111.352 ms	6 - 217 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	57.922 ms	6 - 8 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® SA8775P	58.439 ms	6 - 218 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS9075	61.043 ms	6 - 14 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	63.916 ms	6 - 270 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® SA7255P	111.352 ms	6 - 217 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® SA8295P	66.554 ms	6 - 220 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	41.33 ms	6 - 240 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	47.407 ms	6 - 271 MB	NPU
PidNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	27.003 ms	1 - 337 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS8275 (Proxy)	121.07 ms	3 - 232 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	39.328 ms	2 - 5 MB	NPU
PidNet	TFLITE	float	Qualcomm® SA8775P	48.171 ms	0 - 231 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS9075	61.649 ms	0 - 45 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS8450 (Proxy)	76.087 ms	3 - 346 MB	NPU
PidNet	TFLITE	float	Qualcomm® SA7255P	121.07 ms	3 - 232 MB	NPU
PidNet	TFLITE	float	Qualcomm® SA8295P	53.491 ms	2 - 244 MB	NPU
PidNet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	19.33 ms	1 - 255 MB	NPU
PidNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	12.463 ms	2 - 282 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	38.156 ms	1 - 269 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS6490	206.238 ms	3 - 72 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	98.717 ms	1 - 213 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	50.434 ms	0 - 2 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® SA8775P	51.134 ms	1 - 213 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS9075	53.258 ms	0 - 16 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCM6690	223.507 ms	2 - 233 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	58.793 ms	1 - 270 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® SA7255P	98.717 ms	1 - 213 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® SA8295P	58.192 ms	0 - 215 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	38.304 ms	1 - 233 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	77.878 ms	1 - 221 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	45.649 ms	0 - 263 MB	NPU

License

The license for the original implementation of PidNet can be found here.

References

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.

Downloads last month: -; Downloads are not tracked for this model. How to track

Paper for qualcomm/PidNet

PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers

Paper • 2206.02066 • Published Jun 4, 2022