DDRNet23-Slim: Optimized for Qualcomm Devices

DDRNet23Slim is a machine learning model that segments an image into semantic classes, specifically designed for road-based scenes. It is designed for the application of self-driving cars.

This is based on the implementation of DDRNet23-Slim 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 DDRNet23-Slim 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 DDRNet23-Slim on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

Model checkpoint: DDRNet23s_imagenet.pth
Inference latency: RealTime
Input resolution: 2048x1024
Number of output classes: 19
Number of parameters: 6.13M
Model size (float): 21.7 MB
Model size (w8a8): 6.11 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
DDRNet23-Slim	ONNX	float	Snapdragon® X Elite	27.992 ms	24 - 24 MB	NPU
DDRNet23-Slim	ONNX	float	Snapdragon® 8 Gen 3 Mobile	19.939 ms	31 - 307 MB	NPU
DDRNet23-Slim	ONNX	float	Qualcomm® QCS8550 (Proxy)	29.209 ms	24 - 27 MB	NPU
DDRNet23-Slim	ONNX	float	Qualcomm® QCS9075	39.569 ms	24 - 51 MB	NPU
DDRNet23-Slim	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	13.57 ms	7 - 205 MB	NPU
DDRNet23-Slim	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	10.439 ms	29 - 254 MB	NPU
DDRNet23-Slim	ONNX	float	Snapdragon® X2 Elite	10.917 ms	22 - 22 MB	NPU
DDRNet23-Slim	ONNX	w8a8	Snapdragon® X Elite	91.381 ms	109 - 109 MB	NPU
DDRNet23-Slim	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	43.734 ms	91 - 339 MB	NPU
DDRNet23-Slim	ONNX	w8a8	Qualcomm® QCS6490	300.152 ms	198 - 215 MB	CPU
DDRNet23-Slim	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	58.342 ms	86 - 89 MB	NPU
DDRNet23-Slim	ONNX	w8a8	Qualcomm® QCS9075	63.461 ms	87 - 89 MB	NPU
DDRNet23-Slim	ONNX	w8a8	Qualcomm® QCM6690	265.799 ms	132 - 141 MB	CPU
DDRNet23-Slim	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	42.623 ms	81 - 270 MB	NPU
DDRNet23-Slim	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	250.656 ms	138 - 147 MB	CPU
DDRNet23-Slim	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	44.146 ms	63 - 255 MB	NPU
DDRNet23-Slim	ONNX	w8a8	Snapdragon® X2 Elite	44.16 ms	109 - 109 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Snapdragon® X Elite	33.542 ms	24 - 24 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	22.415 ms	23 - 306 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	98.342 ms	24 - 222 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	32.838 ms	24 - 50 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Qualcomm® SA8775P	40.297 ms	24 - 223 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Qualcomm® QCS9075	53.936 ms	24 - 52 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	67.372 ms	5 - 289 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Qualcomm® SA7255P	98.342 ms	24 - 222 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Qualcomm® SA8295P	42.941 ms	24 - 230 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	15.426 ms	23 - 246 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	10.384 ms	24 - 260 MB	NPU
DDRNet23-Slim	QNN_DLC	float	Snapdragon® X2 Elite	11.701 ms	24 - 24 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Snapdragon® X Elite	58.678 ms	6 - 6 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	42.092 ms	6 - 256 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	107.77 ms	6 - 205 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	56.223 ms	6 - 20 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Qualcomm® SA8775P	56.882 ms	6 - 205 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Qualcomm® QCS9075	59.965 ms	6 - 14 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	61.386 ms	6 - 254 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Qualcomm® SA7255P	107.77 ms	6 - 205 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Qualcomm® SA8295P	64.376 ms	6 - 208 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	40.501 ms	6 - 222 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	46.914 ms	6 - 239 MB	NPU
DDRNet23-Slim	QNN_DLC	w8a8	Snapdragon® X2 Elite	47.663 ms	6 - 6 MB	NPU
DDRNet23-Slim	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	22.471 ms	2 - 295 MB	NPU
DDRNet23-Slim	TFLITE	float	Qualcomm® QCS8275 (Proxy)	98.342 ms	2 - 205 MB	NPU
DDRNet23-Slim	TFLITE	float	Qualcomm® QCS8550 (Proxy)	32.709 ms	2 - 225 MB	NPU
DDRNet23-Slim	TFLITE	float	Qualcomm® SA8775P	185.367 ms	2 - 206 MB	NPU
DDRNet23-Slim	TFLITE	float	Qualcomm® QCS9075	53.574 ms	0 - 40 MB	NPU
DDRNet23-Slim	TFLITE	float	Qualcomm® QCS8450 (Proxy)	67.254 ms	2 - 297 MB	NPU
DDRNet23-Slim	TFLITE	float	Qualcomm® SA7255P	98.342 ms	2 - 205 MB	NPU
DDRNet23-Slim	TFLITE	float	Qualcomm® SA8295P	43.032 ms	2 - 216 MB	NPU
DDRNet23-Slim	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	15.326 ms	1 - 226 MB	NPU
DDRNet23-Slim	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	10.34 ms	2 - 242 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	37.113 ms	0 - 253 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® QCS6490	195.613 ms	11 - 79 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	95.161 ms	1 - 198 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	48.913 ms	1 - 3 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® SA8775P	49.556 ms	1 - 198 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® QCS9075	51.609 ms	1 - 15 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® QCM6690	215.731 ms	10 - 196 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	56.469 ms	0 - 250 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® SA7255P	95.161 ms	1 - 198 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Qualcomm® SA8295P	56.025 ms	1 - 202 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	65.413 ms	0 - 218 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	66.977 ms	9 - 209 MB	NPU
DDRNet23-Slim	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	44.674 ms	1 - 231 MB	NPU

License

The license for the original implementation of DDRNet23-Slim 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/DDRNet23-Slim

Deep Dual-resolution Networks for Real-time and Accurate Semantic Segmentation of Road Scenes

Paper • 2101.06085 • Published Jan 15, 2021