ResNeXt101: Optimized for Qualcomm Devices

ResNeXt101 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 ResNeXt101 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 ResNeXt101 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 ResNeXt101 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: 88.7M
• Model size (float): 338 MB
• Model size (w8a8): 87.3 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
ResNeXt101	ONNX	float	Snapdragon® X2 Elite	3.091 ms	173 - 173 MB	NPU
ResNeXt101	ONNX	float	Snapdragon® X Elite	6.718 ms	172 - 172 MB	NPU
ResNeXt101	ONNX	float	Snapdragon® 8 Gen 3 Mobile	4.584 ms	0 - 383 MB	NPU
ResNeXt101	ONNX	float	Qualcomm® QCS8550 (Proxy)	6.651 ms	1 - 4 MB	NPU
ResNeXt101	ONNX	float	Qualcomm® QCS9075	9.735 ms	0 - 4 MB	NPU
ResNeXt101	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	3.904 ms	0 - 174 MB	NPU
ResNeXt101	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	3.043 ms	1 - 203 MB	NPU
ResNeXt101	ONNX	w8a8	Snapdragon® X2 Elite	1.395 ms	87 - 87 MB	NPU
ResNeXt101	ONNX	w8a8	Snapdragon® X Elite	3.123 ms	87 - 87 MB	NPU
ResNeXt101	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	2.136 ms	0 - 255 MB	NPU
ResNeXt101	ONNX	w8a8	Qualcomm® QCS6490	111.944 ms	8 - 42 MB	CPU
ResNeXt101	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	2.911 ms	0 - 102 MB	NPU
ResNeXt101	ONNX	w8a8	Qualcomm® QCS9075	3.131 ms	0 - 3 MB	NPU
ResNeXt101	ONNX	w8a8	Qualcomm® QCM6690	73.196 ms	1 - 12 MB	CPU
ResNeXt101	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.854 ms	0 - 220 MB	NPU
ResNeXt101	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	68.935 ms	0 - 12 MB	CPU
ResNeXt101	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.574 ms	0 - 216 MB	NPU
ResNeXt101	QNN_DLC	float	Snapdragon® X2 Elite	3.731 ms	1 - 1 MB	NPU
ResNeXt101	QNN_DLC	float	Snapdragon® X Elite	6.915 ms	1 - 1 MB	NPU
ResNeXt101	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	4.62 ms	0 - 367 MB	NPU
ResNeXt101	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	35.889 ms	1 - 191 MB	NPU
ResNeXt101	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	6.978 ms	1 - 3 MB	NPU
ResNeXt101	QNN_DLC	float	Qualcomm® SA8775P	10.322 ms	1 - 204 MB	NPU
ResNeXt101	QNN_DLC	float	Qualcomm® QCS9075	9.992 ms	3 - 5 MB	NPU
ResNeXt101	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	9.859 ms	0 - 313 MB	NPU
ResNeXt101	QNN_DLC	float	Qualcomm® SA7255P	35.889 ms	1 - 191 MB	NPU
ResNeXt101	QNN_DLC	float	Qualcomm® SA8295P	10.901 ms	1 - 140 MB	NPU
ResNeXt101	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	4.036 ms	1 - 189 MB	NPU
ResNeXt101	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	3.089 ms	1 - 193 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Snapdragon® X2 Elite	1.709 ms	0 - 0 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Snapdragon® X Elite	3.068 ms	0 - 0 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	2.128 ms	0 - 250 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® QCS6490	9.07 ms	0 - 2 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	6.549 ms	0 - 206 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	2.816 ms	0 - 2 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® SA8775P	3.505 ms	0 - 207 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® QCS9075	3.133 ms	0 - 2 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® QCM6690	26.9 ms	0 - 367 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	4.027 ms	0 - 253 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® SA7255P	6.549 ms	0 - 206 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Qualcomm® SA8295P	4.306 ms	0 - 209 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.803 ms	0 - 204 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	4.068 ms	0 - 239 MB	NPU
ResNeXt101	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.52 ms	0 - 210 MB	NPU
ResNeXt101	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	4.688 ms	0 - 580 MB	NPU
ResNeXt101	TFLITE	float	Qualcomm® QCS8275 (Proxy)	35.88 ms	0 - 396 MB	NPU
ResNeXt101	TFLITE	float	Qualcomm® QCS8550 (Proxy)	6.834 ms	0 - 2 MB	NPU
ResNeXt101	TFLITE	float	Qualcomm® SA8775P	10.321 ms	0 - 395 MB	NPU
ResNeXt101	TFLITE	float	Qualcomm® QCS9075	10.041 ms	0 - 174 MB	NPU
ResNeXt101	TFLITE	float	Qualcomm® QCS8450 (Proxy)	9.941 ms	0 - 511 MB	NPU
ResNeXt101	TFLITE	float	Qualcomm® SA7255P	35.88 ms	0 - 396 MB	NPU
ResNeXt101	TFLITE	float	Qualcomm® SA8295P	10.94 ms	0 - 336 MB	NPU
ResNeXt101	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	3.974 ms	0 - 389 MB	NPU
ResNeXt101	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	3.091 ms	0 - 401 MB	NPU
ResNeXt101	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	1.977 ms	0 - 250 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® QCS6490	8.901 ms	0 - 88 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	6.289 ms	0 - 209 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	2.658 ms	0 - 2 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® SA8775P	12.696 ms	0 - 210 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® QCS9075	2.95 ms	0 - 89 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® QCM6690	24.89 ms	0 - 371 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	3.966 ms	0 - 249 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® SA7255P	6.289 ms	0 - 209 MB	NPU
ResNeXt101	TFLITE	w8a8	Qualcomm® SA8295P	4.112 ms	0 - 213 MB	NPU
ResNeXt101	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	1.784 ms	0 - 211 MB	NPU
ResNeXt101	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	3.948 ms	0 - 225 MB	NPU
ResNeXt101	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.473 ms	0 - 210 MB	NPU

License

• The license for the original implementation of ResNeXt101 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.