LaMa-Dilated: Optimized for Qualcomm Devices

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

This is based on the implementation of LaMa-Dilated 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 LaMa-Dilated 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 LaMa-Dilated on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_editing

Model Stats:

• Model checkpoint: Dilated CelebAHQ
• Input resolution: 512x512
• Number of parameters: 45.6M
• Model size (float): 174 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
LaMa-Dilated	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	34.496 ms	0 - 389 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® X2 Elite	36.307 ms	90 - 90 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® X Elite	78.171 ms	89 - 89 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® 8 Gen 3 Mobile	54.212 ms	0 - 488 MB	NPU
LaMa-Dilated	ONNX	float	Qualcomm® QCS8550 (Proxy)	76.899 ms	10 - 14 MB	NPU
LaMa-Dilated	ONNX	float	Qualcomm® QCS9075	117.124 ms	6 - 13 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	41.78 ms	7 - 418 MB	NPU
LaMa-Dilated	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	79.984 ms	59 - 695 MB	NPU
LaMa-Dilated	ONNX	w8a16	Snapdragon® X2 Elite	88.353 ms	68 - 68 MB	NPU
LaMa-Dilated	ONNX	w8a16	Snapdragon® X Elite	165.854 ms	68 - 68 MB	NPU
LaMa-Dilated	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	110.879 ms	54 - 704 MB	NPU
LaMa-Dilated	ONNX	w8a16	Qualcomm® QCS6490	10027.083 ms	423 - 430 MB	CPU
LaMa-Dilated	ONNX	w8a16	Qualcomm® QCM6690	4495.056 ms	369 - 380 MB	CPU
LaMa-Dilated	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	92.13 ms	58 - 588 MB	NPU
LaMa-Dilated	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	4512.583 ms	359 - 370 MB	CPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	32.35 ms	4 - 390 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® X2 Elite	34.445 ms	4 - 4 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® X Elite	75.642 ms	4 - 4 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	52.301 ms	3 - 473 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	403.464 ms	1 - 360 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	74.921 ms	4 - 16 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® SA8775P	107.144 ms	1 - 360 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS9075	120.064 ms	4 - 13 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	137.783 ms	2 - 417 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® SA7255P	403.464 ms	1 - 360 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® SA8295P	113.822 ms	1 - 342 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	40.383 ms	1 - 363 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	27.902 ms	1 - 765 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Snapdragon® X2 Elite	30.558 ms	2 - 2 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Snapdragon® X Elite	66.716 ms	2 - 2 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	44.225 ms	0 - 760 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Qualcomm® QCS8275 (Proxy)	132.83 ms	2 - 613 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	64.914 ms	2 - 4 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Qualcomm® SA8775P	65.396 ms	2 - 614 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Qualcomm® QCS9075	79.422 ms	1 - 6 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Qualcomm® QCS8450 (Proxy)	118.108 ms	1 - 577 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Qualcomm® SA7255P	132.83 ms	2 - 613 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Qualcomm® SA8295P	90.45 ms	2 - 434 MB	NPU
LaMa-Dilated	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	32.483 ms	2 - 614 MB	NPU
LaMa-Dilated	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	32.741 ms	2 - 400 MB	NPU
LaMa-Dilated	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	53.069 ms	1 - 549 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS8275 (Proxy)	404.172 ms	3 - 375 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS8550 (Proxy)	74.323 ms	3 - 132 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA8775P	106.24 ms	0 - 371 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS9075	119.362 ms	2 - 104 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS8450 (Proxy)	136.127 ms	3 - 495 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA7255P	404.172 ms	3 - 375 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA8295P	113.852 ms	3 - 357 MB	NPU
LaMa-Dilated	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	40.461 ms	2 - 380 MB	NPU

License

• The license for the original implementation of LaMa-Dilated can be found here.

References

• Resolution-robust Large Mask Inpainting with Fourier Convolutions
• 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.