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 1.24.3	Download
ONNX	w8a16	Universal	QAIRT 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	float	Universal	QAIRT 2.43	Download
QNN_DLC	w8a16	Universal	QAIRT 2.43	Download
TFLITE	float	Universal	QAIRT 2.43, TFLite 2.19.1	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® 8 Elite Gen 5 Mobile	59.145 ms	10 - 504 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® X2 Elite	60.821 ms	32 - 32 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® X Elite	149.494 ms	31 - 31 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® 8 Gen 3 Mobile	102.354 ms	0 - 742 MB	NPU
AOT-GAN	ONNX	float	Qualcomm® QCS8550 (Proxy)	145.382 ms	0 - 39 MB	NPU
AOT-GAN	ONNX	float	Qualcomm® QCS9075	225.257 ms	4 - 11 MB	NPU
AOT-GAN	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	78.4 ms	7 - 635 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	35.15 ms	5 - 616 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® X2 Elite	37.834 ms	23 - 23 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® X Elite	85.675 ms	21 - 21 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	61.53 ms	5 - 716 MB	NPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCS6490	13222.891 ms	395 - 398 MB	CPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	82.141 ms	0 - 27 MB	NPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCS9075	109.453 ms	5 - 8 MB	NPU
AOT-GAN	ONNX	w8a16	Qualcomm® QCM6690	6708.473 ms	326 - 334 MB	CPU
AOT-GAN	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	46.974 ms	5 - 565 MB	NPU
AOT-GAN	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	6582.488 ms	332 - 341 MB	CPU
AOT-GAN	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	48.088 ms	3 - 489 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® X2 Elite	52.381 ms	4 - 4 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® X Elite	125.294 ms	4 - 4 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	88.961 ms	4 - 728 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	544.139 ms	1 - 540 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	124.665 ms	4 - 7 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® SA8775P	164.745 ms	1 - 540 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS9075	213.598 ms	4 - 13 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	201.313 ms	3 - 646 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® SA7255P	544.139 ms	1 - 540 MB	NPU
AOT-GAN	QNN_DLC	float	Qualcomm® SA8295P	179.741 ms	1 - 478 MB	NPU
AOT-GAN	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	70.983 ms	3 - 579 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	35.651 ms	2 - 628 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® X2 Elite	62.116 ms	2 - 2 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® X Elite	84.345 ms	2 - 2 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	62.458 ms	0 - 718 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS6490	298.545 ms	3 - 8 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS8275 (Proxy)	189.995 ms	2 - 511 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	82.451 ms	2 - 5 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® SA8775P	81.984 ms	2 - 513 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS9075	110.242 ms	2 - 7 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCM6690	1239.264 ms	2 - 673 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® QCS8450 (Proxy)	144.92 ms	2 - 711 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® SA7255P	189.995 ms	2 - 511 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Qualcomm® SA8295P	115.655 ms	2 - 511 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	48.399 ms	2 - 563 MB	NPU
AOT-GAN	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	124.32 ms	2 - 672 MB	NPU
AOT-GAN	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	47.463 ms	3 - 504 MB	NPU
AOT-GAN	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	88.948 ms	3 - 764 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS8275 (Proxy)	544.417 ms	3 - 561 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS8550 (Proxy)	124.676 ms	3 - 6 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® SA8775P	163.958 ms	0 - 557 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS9075	210.896 ms	2 - 45 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® QCS8450 (Proxy)	200.805 ms	4 - 679 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® SA7255P	544.417 ms	3 - 561 MB	NPU
AOT-GAN	TFLITE	float	Qualcomm® SA8295P	178.998 ms	0 - 487 MB	NPU
AOT-GAN	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	70.96 ms	2 - 597 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.