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 2.42, 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 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® X2 Elite	14.154 ms	22 - 22 MB	NPU
PidNet	ONNX	float	Snapdragon® X Elite	34.21 ms	24 - 24 MB	NPU
PidNet	ONNX	float	Snapdragon® 8 Gen 3 Mobile	24.114 ms	18 - 334 MB	NPU
PidNet	ONNX	float	Qualcomm® QCS8550 (Proxy)	34.459 ms	24 - 27 MB	NPU
PidNet	ONNX	float	Qualcomm® QCS9075	47.725 ms	24 - 51 MB	NPU
PidNet	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	17.416 ms	6 - 225 MB	NPU
PidNet	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	13.39 ms	30 - 298 MB	NPU
PidNet	ONNX	w8a8	Snapdragon® X2 Elite	45.505 ms	134 - 134 MB	NPU
PidNet	ONNX	w8a8	Snapdragon® X Elite	85.169 ms	133 - 133 MB	NPU
PidNet	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	45.893 ms	109 - 371 MB	NPU
PidNet	ONNX	w8a8	Qualcomm® QCS6490	395.24 ms	197 - 215 MB	CPU
PidNet	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	60.084 ms	101 - 107 MB	NPU
PidNet	ONNX	w8a8	Qualcomm® QCS9075	66.844 ms	104 - 106 MB	NPU
PidNet	ONNX	w8a8	Qualcomm® QCM6690	348.38 ms	198 - 208 MB	CPU
PidNet	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	43.624 ms	98 - 308 MB	NPU
PidNet	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	327.876 ms	211 - 222 MB	CPU
PidNet	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	45.169 ms	99 - 319 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® X2 Elite	13.818 ms	24 - 24 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® X Elite	40.239 ms	24 - 24 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	26.914 ms	24 - 343 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	121.006 ms	24 - 244 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	38.712 ms	24 - 36 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® SA8775P	48.083 ms	24 - 244 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS9075	61.877 ms	24 - 52 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	76.2 ms	5 - 336 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® SA7255P	121.006 ms	24 - 244 MB	NPU
PidNet	QNN_DLC	float	Qualcomm® SA8295P	52.875 ms	24 - 261 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	19.401 ms	21 - 266 MB	NPU
PidNet	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	12.273 ms	10 - 281 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® X2 Elite	48.221 ms	6 - 6 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® X Elite	60.443 ms	6 - 6 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	43.161 ms	6 - 272 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	111.117 ms	6 - 217 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	58.012 ms	6 - 8 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® SA8775P	58.495 ms	6 - 218 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS9075	61.544 ms	6 - 14 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	64.308 ms	6 - 271 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® SA7255P	111.117 ms	6 - 217 MB	NPU
PidNet	QNN_DLC	w8a8	Qualcomm® SA8295P	66.512 ms	6 - 220 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	41.45 ms	6 - 240 MB	NPU
PidNet	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	47.754 ms	6 - 271 MB	NPU
PidNet	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	26.686 ms	1 - 335 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS8275 (Proxy)	121.027 ms	2 - 232 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS8550 (Proxy)	38.953 ms	2 - 54 MB	NPU
PidNet	TFLITE	float	Qualcomm® SA8775P	48.051 ms	2 - 233 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS9075	62.137 ms	0 - 45 MB	NPU
PidNet	TFLITE	float	Qualcomm® QCS8450 (Proxy)	77.886 ms	3 - 344 MB	NPU
PidNet	TFLITE	float	Qualcomm® SA7255P	121.027 ms	2 - 232 MB	NPU
PidNet	TFLITE	float	Qualcomm® SA8295P	52.951 ms	2 - 244 MB	NPU
PidNet	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	19.491 ms	2 - 251 MB	NPU
PidNet	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	12.232 ms	2 - 276 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	37.912 ms	1 - 270 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS6490	206.978 ms	3 - 73 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	98.64 ms	0 - 211 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	50.438 ms	0 - 167 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® SA8775P	51.148 ms	1 - 213 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS9075	53.612 ms	1 - 17 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCM6690	238.637 ms	3 - 197 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	58.989 ms	1 - 267 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® SA7255P	98.64 ms	0 - 211 MB	NPU
PidNet	TFLITE	w8a8	Qualcomm® SA8295P	58.249 ms	1 - 215 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	38.413 ms	0 - 234 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	66.975 ms	2 - 221 MB	NPU
PidNet	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	41.186 ms	1 - 265 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