ResNet-2Plus1D: Optimized for Qualcomm Devices

ResNet (2+1)D Convolutions is a network which explicitly factorizes 3D convolution into two separate and successive operations, a 2D spatial convolution and a 1D temporal convolution. It used for video understanding applications.

This is based on the implementation of ResNet-2Plus1D 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 1.24.3	Download
ONNX	w8a8	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a8	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download
TFLITE	w8a8	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit ResNet-2Plus1D 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 ResNet-2Plus1D on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.video_classification

Model Stats:

• Model checkpoint: Kinetics-400
• Input resolution: 112x112
• Number of parameters: 31.5M
• Model size (float): 120 MB
• Model size (w8a8): 30.8 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
ResNet-2Plus1D	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	5.523 ms	2 - 214 MB	NPU
ResNet-2Plus1D	ONNX	float	Snapdragon® 8 Elite Mobile	7.215 ms	0 - 209 MB	NPU
ResNet-2Plus1D	ONNX	float	Snapdragon® X2 Elite	6.145 ms	60 - 60 MB	NPU
ResNet-2Plus1D	ONNX	float	Snapdragon® X Elite	12.211 ms	60 - 60 MB	NPU
ResNet-2Plus1D	ONNX	float	Snapdragon® X Elite	12.211 ms	60 - 60 MB	NPU
ResNet-2Plus1D	ONNX	float	Snapdragon® 8 Gen 3 Mobile	9.079 ms	2 - 299 MB	NPU
ResNet-2Plus1D	ONNX	float	Qualcomm® QCS8550 (Proxy)	12.028 ms	2 - 5 MB	NPU
ResNet-2Plus1D	ONNX	float	Qualcomm® QCS9075	21.357 ms	2 - 7 MB	NPU
ResNet-2Plus1D	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	7.215 ms	0 - 209 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.923 ms	0 - 191 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® 8 Elite Mobile	2.628 ms	0 - 189 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® X2 Elite	2.039 ms	31 - 31 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® X Elite	4.546 ms	31 - 31 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® X Elite	4.546 ms	31 - 31 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® 8 Gen 3 Mobile	3.267 ms	0 - 224 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Qualcomm® QCS6490	317.614 ms	97 - 127 MB	CPU
ResNet-2Plus1D	ONNX	w8a8	Qualcomm® QCS8550 (Proxy)	4.274 ms	0 - 227 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Qualcomm® QCS9075	4.11 ms	1 - 3 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Qualcomm® QCM6690	301.447 ms	100 - 108 MB	CPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	2.628 ms	0 - 189 MB	NPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	269.891 ms	67 - 75 MB	CPU
ResNet-2Plus1D	ONNX	w8a8	Snapdragon® 7 Gen 4 Mobile	269.891 ms	67 - 75 MB	CPU
ResNet-2Plus1D	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	5.641 ms	2 - 227 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Snapdragon® 8 Elite Mobile	7.17 ms	0 - 217 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Snapdragon® X2 Elite	6.646 ms	2 - 2 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Snapdragon® X Elite	12.87 ms	2 - 2 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Snapdragon® X Elite	12.87 ms	2 - 2 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	9.25 ms	0 - 291 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	81.878 ms	2 - 215 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	12.631 ms	2 - 4 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® SA8775P	21.262 ms	1 - 215 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® SA8775P	21.262 ms	1 - 215 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® SA8775P	21.262 ms	1 - 215 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® QCS9075	23.466 ms	2 - 6 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	29.007 ms	0 - 268 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® SA7255P	81.878 ms	2 - 215 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Qualcomm® SA8295P	22.738 ms	0 - 197 MB	NPU
ResNet-2Plus1D	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	7.17 ms	0 - 217 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	1.849 ms	0 - 184 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® 8 Elite Mobile	2.565 ms	0 - 182 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® X2 Elite	2.366 ms	1 - 1 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® X Elite	4.906 ms	1 - 1 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® X Elite	4.906 ms	1 - 1 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® 8 Gen 3 Mobile	3.339 ms	1 - 219 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® QCS6490	19.672 ms	3 - 5 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® QCS8275 (Proxy)	13.385 ms	1 - 183 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® QCS8550 (Proxy)	4.599 ms	1 - 3 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® SA8775P	4.698 ms	1 - 187 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® SA8775P	4.698 ms	1 - 187 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® SA8775P	4.698 ms	1 - 187 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® QCS9075	4.76 ms	1 - 3 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® QCM6690	82.143 ms	1 - 198 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® QCS8450 (Proxy)	7.861 ms	1 - 217 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® SA7255P	13.385 ms	1 - 183 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Qualcomm® SA8295P	7.843 ms	0 - 180 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	2.565 ms	0 - 182 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	7.895 ms	1 - 189 MB	NPU
ResNet-2Plus1D	QNN_DLC	w8a8	Snapdragon® 7 Gen 4 Mobile	7.895 ms	1 - 189 MB	NPU
ResNet-2Plus1D	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	257.361 ms	0 - 247 MB	NPU
ResNet-2Plus1D	TFLITE	float	Snapdragon® 8 Elite Mobile	256.547 ms	0 - 242 MB	NPU
ResNet-2Plus1D	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	294.591 ms	0 - 305 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® QCS8275 (Proxy)	714.781 ms	0 - 237 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® QCS8550 (Proxy)	374.47 ms	0 - 2 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® SA8775P	392.912 ms	0 - 235 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® SA8775P	392.912 ms	0 - 235 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® SA8775P	392.912 ms	0 - 235 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® QCS9075	390.284 ms	0 - 66 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® QCS8450 (Proxy)	429.829 ms	0 - 301 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® SA7255P	714.781 ms	0 - 237 MB	NPU
ResNet-2Plus1D	TFLITE	float	Qualcomm® SA8295P	453.705 ms	0 - 228 MB	NPU
ResNet-2Plus1D	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	256.547 ms	0 - 242 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Snapdragon® 8 Elite Gen 5 Mobile	643.395 ms	0 - 454 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Snapdragon® 8 Elite Mobile	507.9 ms	0 - 546 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Snapdragon® 8 Gen 3 Mobile	586.817 ms	0 - 516 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® QCS6490	1781.787 ms	267 - 428 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® QCS8275 (Proxy)	1517.615 ms	0 - 438 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® QCS8550 (Proxy)	751.095 ms	0 - 3 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® SA8775P	799.258 ms	0 - 440 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® SA8775P	799.258 ms	0 - 440 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® SA8775P	799.258 ms	0 - 440 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® QCS9075	573.098 ms	0 - 65 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® QCM6690	1602.624 ms	296 - 458 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® QCS8450 (Proxy)	866.104 ms	0 - 463 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® SA7255P	1517.615 ms	0 - 438 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Qualcomm® SA8295P	871.702 ms	0 - 435 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Snapdragon® 8 Elite For Galaxy Mobile	507.9 ms	0 - 546 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	1215.928 ms	295 - 365 MB	NPU
ResNet-2Plus1D	TFLITE	w8a8	Snapdragon® 7 Gen 4 Mobile	1215.928 ms	295 - 365 MB	NPU

License

• The license for the original implementation of ResNet-2Plus1D can be found here.

References

• A Closer Look at Spatiotemporal Convolutions for Action Recognition
• 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.