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 2.45, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	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® X2 Elite	36.379 ms	209 - 209 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® X Elite	77.542 ms	145 - 145 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® 8 Gen 3 Mobile	55.147 ms	4 - 474 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® 8 Gen 1 Mobile	134.424 ms	11 - 528 MB	NPU
LaMa-Dilated	ONNX	float	Qualcomm® QCS8550 (Proxy)	76.598 ms	0 - 96 MB	NPU
LaMa-Dilated	ONNX	float	Qualcomm® QCS8450	134.424 ms	11 - 528 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® 8 Elite Mobile	43.587 ms	9 - 439 MB	NPU
LaMa-Dilated	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	34.063 ms	7 - 400 MB	NPU
LaMa-Dilated	ONNX	float	Qualcomm® QCS9075	116.421 ms	7 - 52 MB	NPU
LaMa-Dilated	ONNX	float	Qualcomm® QCS8750	43.587 ms	9 - 439 MB	NPU
LaMa-Dilated	ONNX	float	Qualcomm® QCS7181	77.542 ms	145 - 145 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® X2 Elite	34.355 ms	4 - 4 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® X Elite	75.448 ms	4 - 4 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	52.661 ms	0 - 451 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	133.452 ms	0 - 405 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS8275	403.201 ms	1 - 364 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	75.939 ms	4 - 6 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS8450	133.452 ms	0 - 405 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® 8 Elite Mobile	40.229 ms	1 - 367 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® SA8295P	109.537 ms	1 - 299 MB	NPU
LaMa-Dilated	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	32.231 ms	4 - 397 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® SA7255P	403.201 ms	1 - 364 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS9075	118.107 ms	4 - 13 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS8750	40.229 ms	1 - 367 MB	NPU
LaMa-Dilated	QNN_DLC	float	Qualcomm® QCS7181	75.448 ms	4 - 4 MB	NPU
LaMa-Dilated	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	53.344 ms	3 - 546 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS8275	404.436 ms	3 - 381 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS8550 (Proxy)	75.467 ms	3 - 132 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA8775P	1398.545 ms	0 - 24 MB	GPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA8650P	1398.545 ms	0 - 24 MB	GPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA8255P	1398.545 ms	0 - 24 MB	GPU
LaMa-Dilated	TFLITE	float	Snapdragon® 8 Elite Mobile	40.418 ms	1 - 391 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA8295P	110.134 ms	3 - 311 MB	NPU
LaMa-Dilated	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	32.274 ms	2 - 416 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® SA7255P	404.436 ms	3 - 381 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS9075	118.829 ms	2 - 104 MB	NPU
LaMa-Dilated	TFLITE	float	Qualcomm® QCS8750	40.418 ms	1 - 391 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.