AOT-GAN: Optimized for Qualcomm Devices

AOT-GAN is a machine learning model that allows to erase and in-paint part of given input image.

This is based on the implementation of AOT-GAN 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	w8a16	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a16	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.17.0	Download

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

Model Details

Model Type: Model_use_case.image_editing

Model Stats:

• Model checkpoint: CelebAHQ
• Input resolution: 512x512
• Number of parameters: 15.2M
• Model size (float): 58.0 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
AOT-GAN	ONNX	float	Snapdragon® X2 Elite	61.291 ms	32 - 32 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® X Elite	149.23 ms	31 - 31 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® 8 Gen 3 Mobile	102.836 ms	11 - 762 MB	NPU
AOT-GAN	ONNX	float	Qualcomm® QCS8550 (Proxy)	145.095 ms	0 - 42 MB	NPU
AOT-GAN	ONNX	float	Qualcomm® QCS9075	225.942 ms	4 - 11 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	78.222 ms	9 - 636 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	58.242 ms	11 - 504 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® X2 Elite	37.821 ms	23 - 23 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® X Elite	85.628 ms	21 - 21 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	61.821 ms	5 - 723 MB	NPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCS6490	13218.859 ms	394 - 396 MB	CPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	82.195 ms	0 - 25 MB	NPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCS9075	109.763 ms	5 - 8 MB	NPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCM6690	6558.865 ms	326 - 334 MB	CPU
AOT-GAN	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	47.107 ms	5 - 526 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	6568.009 ms	333 - 341 MB	CPU
AOT-GAN	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	34.817 ms	5 - 622 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® X2 Elite	52.604 ms	4 - 4 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® X Elite	125.797 ms	4 - 4 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	89.365 ms	2 - 726 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	545.515 ms	1 - 540 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	124.651 ms	4 - 732 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® SA8775P	164.958 ms	1 - 540 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS9075	214.543 ms	4 - 13 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	202.149 ms	3 - 645 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® SA7255P	545.515 ms	1 - 540 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® SA8295P	179.861 ms	1 - 479 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	70.788 ms	1 - 580 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	47.377 ms	4 - 490 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® X2 Elite	39.686 ms	2 - 2 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® X Elite	84.221 ms	2 - 2 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	62.909 ms	1 - 718 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS6490	298.491 ms	1 - 6 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS8275 (Proxy)	190.225 ms	2 - 511 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	82.536 ms	2 - 4 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® SA8775P	386.756 ms	2 - 510 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS9075	109.788 ms	2 - 7 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCM6690	1325.187 ms	2 - 666 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS8450 (Proxy)	146.613 ms	2 - 717 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® SA7255P	190.225 ms	2 - 511 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® SA8295P	115.704 ms	2 - 510 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	48.174 ms	2 - 564 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	124.583 ms	2 - 671 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	35.651 ms	2 - 628 MB	NPU
AOT-GAN	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	88.814 ms	3 - 761 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS8275 (Proxy)	543.046 ms	3 - 559 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS8550 (Proxy)	124.537 ms	3 - 6 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® SA8775P	163.926 ms	3 - 560 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS9075	210.968 ms	2 - 45 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS8450 (Proxy)	201.949 ms	3 - 677 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® SA7255P	543.046 ms	3 - 559 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® SA8295P	179.01 ms	0 - 486 MB	NPU
AOT-GAN	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	70.806 ms	0 - 582 MB	NPU
AOT-GAN	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	46.991 ms	3 - 505 MB	NPU

License

• The license for the original implementation of AOT-GAN can be found here.

References

• Aggregated Contextual Transformations for High-Resolution Image Inpainting
• 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.