You can deploy this model and start earning money today!

qualcomm/
Video-MAE

Name: Video-MAE
Rating: 2.5 (23 reviews)

Views

No views yet

Video-MAE: Optimized for Qualcomm Devices

Video MAE (Masked Auto Encoder) is a network for doing video classification that uses the ViT (Vision Transformer) backbone.

This is based on the implementation of Video-MAE 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 2.45, ONNX Runtime 1.27.1	Download
ONNX	w8a16	Universal	QAIRT 2.45, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a16	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit Video-MAE 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 Video-MAE on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.video_classification

Model Stats:

Input resolution: 224x224
Model checkpoint: Kinectics-400
Model size (float): 335 MB
Number of parameters: 87.7M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Video-MAE	ONNX	float	Snapdragon® X2 Elite	464.527 ms	9 - 9 MB	NPU
Video-MAE	ONNX	float	Snapdragon® X Elite	615.118 ms	188 - 188 MB	NPU
Video-MAE	ONNX	float	Snapdragon® 8 Gen 3 Mobile	398.964 ms	1 - 1086 MB	NPU
Video-MAE	ONNX	float	Snapdragon® 8 Gen 1 Mobile	1389.915 ms	10 - 954 MB	NPU
Video-MAE	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	558.373 ms	9 - 22 MB	NPU
Video-MAE	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	596.743 ms	0 - 217 MB	NPU
Video-MAE	ONNX	float	Qualcomm® QCS8450	1389.915 ms	10 - 954 MB	NPU
Video-MAE	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	561.03 ms	9 - 21 MB	NPU
Video-MAE	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	615.118 ms	188 - 188 MB	NPU
Video-MAE	ONNX	float	Qualcomm® Dragonwing™ Q-8750	379.298 ms	2 - 790 MB	NPU
Video-MAE	ONNX	float	Snapdragon® 8 Elite Mobile	379.298 ms	2 - 790 MB	NPU
Video-MAE	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	469.389 ms	4 - 795 MB	NPU
Video-MAE	ONNX	w8a16	Snapdragon® X2 Elite	329.842 ms	5 - 5 MB	NPU
Video-MAE	ONNX	w8a16	Snapdragon® X Elite	491.778 ms	100 - 100 MB	NPU
Video-MAE	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	369.404 ms	2 - 1092 MB	NPU
Video-MAE	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	1446.026 ms	6 - 1156 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® Dragonwing™ QCS6490	2317.358 ms	0 - 7 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	543.863 ms	1 - 9 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	490.668 ms	0 - 110 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® QCS8450	1446.026 ms	6 - 1156 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	569.515 ms	0 - 7 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	491.778 ms	100 - 100 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	1079.659 ms	4 - 1357 MB	NPU
Video-MAE	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	289.993 ms	1 - 874 MB	NPU
Video-MAE	ONNX	w8a16	Snapdragon® 8 Elite Mobile	289.993 ms	1 - 874 MB	NPU
Video-MAE	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	362.465 ms	1 - 896 MB	NPU
Video-MAE	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	1079.659 ms	4 - 1357 MB	NPU
Video-MAE	QNN_DLC	float	Snapdragon® X2 Elite	254.278 ms	9 - 9 MB	NPU
Video-MAE	QNN_DLC	float	Snapdragon® X Elite	470.333 ms	9 - 9 MB	NPU
Video-MAE	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	382.535 ms	9 - 986 MB	NPU
Video-MAE	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	559.657 ms	9 - 899 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	519.829 ms	9 - 21 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	1125.572 ms	1 - 678 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	460.174 ms	9 - 902 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® SA8775P	480.261 ms	0 - 678 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® SA8650P	480.261 ms	0 - 678 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® SA8255P	480.261 ms	0 - 678 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® QCS8450	559.657 ms	9 - 899 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	475.947 ms	9 - 20 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	470.333 ms	9 - 9 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	266.98 ms	9 - 687 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® SA7255P	1125.572 ms	1 - 678 MB	NPU
Video-MAE	QNN_DLC	float	Qualcomm® SA8295P	616.052 ms	1 - 650 MB	NPU
Video-MAE	QNN_DLC	float	Snapdragon® 8 Elite Mobile	266.98 ms	9 - 687 MB	NPU
Video-MAE	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	290.483 ms	9 - 703 MB	NPU
Video-MAE	QNN_DLC	w8a16	Snapdragon® X2 Elite	335.934 ms	5 - 5 MB	NPU
Video-MAE	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	374.793 ms	5 - 1246 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	518.494 ms	5 - 12 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	980.695 ms	2 - 833 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	506.112 ms	5 - 7 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® SA8775P	514.795 ms	1 - 797 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® SA8650P	514.795 ms	1 - 797 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® SA8255P	514.795 ms	1 - 797 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	509.565 ms	2 - 8 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	3488.789 ms	5 - 1344 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	712.528 ms	5 - 1154 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	291.759 ms	5 - 781 MB	NPU
Video-MAE	QNN_DLC	w8a16	Qualcomm® SA7255P	980.695 ms	2 - 833 MB	NPU
Video-MAE	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	291.759 ms	5 - 781 MB	NPU
Video-MAE	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	334.292 ms	5 - 809 MB	NPU
Video-MAE	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	712.528 ms	5 - 1154 MB	NPU
Video-MAE	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	122.54 ms	0 - 971 MB	NPU
Video-MAE	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	267.562 ms	1 - 935 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	184.668 ms	0 - 208 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	469.689 ms	1 - 682 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	162.658 ms	0 - 3 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® SA8775P	181.324 ms	0 - 710 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® SA8650P	181.324 ms	0 - 710 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® SA8255P	181.324 ms	0 - 710 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® QCS8450	267.562 ms	1 - 935 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	179.46 ms	0 - 207 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	90.397 ms	0 - 685 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® SA7255P	469.689 ms	1 - 682 MB	NPU
Video-MAE	TFLITE	float	Qualcomm® SA8295P	238.617 ms	0 - 649 MB	NPU
Video-MAE	TFLITE	float	Snapdragon® 8 Elite Mobile	90.397 ms	0 - 685 MB	NPU
Video-MAE	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	63.849 ms	0 - 728 MB	NPU

License

The license for the original implementation of Video-MAE 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.

qualcomm/Video-MAE