ResNeXt50: Optimized for Qualcomm Devices

ResNeXt50 is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.

This is based on the implementation of ResNeXt50 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 ResNeXt50 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 ResNeXt50 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_classification

Model Stats:

• Model checkpoint: Imagenet
• Input resolution: 224x224
• Number of parameters: 25.0M
• Model size (float): 95.4 MB
• Model size (w8a8): 24.8 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
ResNeXt50	ONNX	float	Snapdragon® X Elite	2.448 ms	50 - 50 MB	NPU
ResNeXt50	ONNX	float	Snapdragon® 8 Gen 3 Mobile	1.894 ms	0 - 194 MB	NPU
ResNeXt50	ONNX	float	Qualcomm® QCS8550 (Proxy)	2.573 ms	0 - 311 MB	NPU
ResNeXt50	ONNX	float	Qualcomm® QCS9075	3.68 ms	1 - 4 MB	NPU
ResNeXt50	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	1.569 ms	0 - 133 MB	NPU
ResNeXt50	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1.31 ms	0 - 135 MB	NPU
ResNeXt50	ONNX	w8a8	Snapdragon® X Elite	1.236 ms	26 - 26 MB	NPU
ResNeXt50	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	1.051 ms	0 - 168 MB	NPU
ResNeXt50	ONNX	w8a8	Qualcomm® QCS6490	47.343 ms	7 - 29 MB	CPU
ResNeXt50	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	1.345 ms	0 - 30 MB	NPU
ResNeXt50	ONNX	w8a8	Qualcomm® QCS9075	1.483 ms	0 - 3 MB	NPU
ResNeXt50	ONNX	w8a8	Qualcomm® QCM6690	26.039 ms	0 - 9 MB	CPU
ResNeXt50	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.854 ms	0 - 142 MB	NPU
ResNeXt50	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	20.332 ms	0 - 10 MB	CPU
ResNeXt50	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.747 ms	0 - 141 MB	NPU
ResNeXt50	QNN_DLC	float	Snapdragon® X Elite	2.671 ms	1 - 1 MB	NPU
ResNeXt50	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	1.824 ms	0 - 138 MB	NPU
ResNeXt50	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	11.971 ms	1 - 76 MB	NPU
ResNeXt50	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	2.483 ms	1 - 2 MB	NPU
ResNeXt50	QNN_DLC	float	Qualcomm® SA8775P	3.835 ms	1 - 76 MB	NPU
ResNeXt50	QNN_DLC	float	Qualcomm® QCS9075	3.636 ms	1 - 3 MB	NPU
ResNeXt50	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	3.963 ms	0 - 114 MB	NPU
ResNeXt50	QNN_DLC	float	Qualcomm® SA7255P	11.971 ms	1 - 76 MB	NPU
ResNeXt50	QNN_DLC	float	Qualcomm® SA8295P	4.187 ms	0 - 53 MB	NPU
ResNeXt50	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	1.483 ms	0 - 75 MB	NPU
ResNeXt50	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.204 ms	1 - 77 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Snapdragon® X Elite	1.26 ms	0 - 0 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	0.808 ms	0 - 97 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® QCS6490	3.097 ms	0 - 2 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	2.497 ms	0 - 69 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	1.093 ms	0 - 2 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® SA8775P	1.453 ms	0 - 70 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® QCS9075	1.216 ms	2 - 4 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® QCM6690	7.668 ms	0 - 197 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	1.442 ms	0 - 98 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® SA7255P	2.497 ms	0 - 69 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Qualcomm® SA8295P	1.766 ms	0 - 66 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.633 ms	0 - 71 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	1.395 ms	0 - 74 MB	NPU
ResNeXt50	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.523 ms	0 - 72 MB	NPU
ResNeXt50	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	1.784 ms	0 - 187 MB	NPU
ResNeXt50	TFLITE	float	Qualcomm® QCS8275 (Proxy)	12.016 ms	0 - 129 MB	NPU
ResNeXt50	TFLITE	float	Qualcomm® QCS8550 (Proxy)	2.455 ms	0 - 3 MB	NPU
ResNeXt50	TFLITE	float	Qualcomm® SA8775P	3.818 ms	0 - 128 MB	NPU
ResNeXt50	TFLITE	float	Qualcomm® QCS9075	3.684 ms	0 - 52 MB	NPU
ResNeXt50	TFLITE	float	Qualcomm® QCS8450 (Proxy)	4.056 ms	0 - 162 MB	NPU
ResNeXt50	TFLITE	float	Qualcomm® SA7255P	12.016 ms	0 - 129 MB	NPU
ResNeXt50	TFLITE	float	Qualcomm® SA8295P	4.211 ms	0 - 104 MB	NPU
ResNeXt50	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	1.452 ms	0 - 129 MB	NPU
ResNeXt50	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.201 ms	0 - 129 MB	NPU
ResNeXt50	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	0.674 ms	0 - 97 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® QCS6490	2.876 ms	0 - 27 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	2.196 ms	0 - 68 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	0.915 ms	0 - 3 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® SA8775P	4.4 ms	0 - 69 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® QCS9075	1.017 ms	0 - 27 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® QCM6690	7.322 ms	0 - 190 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	1.276 ms	0 - 99 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® SA7255P	2.196 ms	0 - 68 MB	NPU
ResNeXt50	TFLITE	w8a8	Qualcomm® SA8295P	1.558 ms	0 - 62 MB	NPU
ResNeXt50	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	0.557 ms	0 - 73 MB	NPU
ResNeXt50	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	1.239 ms	0 - 66 MB	NPU
ResNeXt50	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	0.48 ms	0 - 71 MB	NPU

License

• The license for the original implementation of ResNeXt50 can be found here.

References

• Aggregated Residual Transformations for Deep Neural Networks
• 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.