DDColor: Optimized for Qualcomm Devices

DDColor is a coloring algorithm that produces natural, vivid color results from incoming black and white images.

This is based on the implementation of DDColor 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 DDColor 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 DDColor on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.image_editing

Model Stats:

• Model checkpoint: ddcolor_paper_tiny.pth
• Input resolution: 224x224
• Number of parameters: 56.3M
• Model size (float): 215 MB
• Model size (w8a8): 54.8 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
DDColor	ONNX	float	Snapdragon® X2 Elite	28.065 ms	212 - 212 MB	NPU
DDColor	ONNX	float	Snapdragon® X Elite	72.637 ms	150 - 150 MB	NPU
DDColor	ONNX	float	Snapdragon® 8 Gen 3 Mobile	45.098 ms	2 - 1468 MB	NPU
DDColor	ONNX	float	Snapdragon® 8 Gen 1 Mobile	103.266 ms	2 - 528 MB	NPU
DDColor	ONNX	float	Qualcomm® QCS8550 (Proxy)	74.158 ms	0 - 135 MB	NPU
DDColor	ONNX	float	Qualcomm® QCS8450	103.266 ms	2 - 528 MB	NPU
DDColor	ONNX	float	Snapdragon® 8 Elite Mobile	35.255 ms	2 - 795 MB	NPU
DDColor	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	27.423 ms	2 - 733 MB	NPU
DDColor	ONNX	float	Qualcomm® QCS9075	86.707 ms	2 - 47 MB	NPU
DDColor	ONNX	float	Qualcomm® QCS8750	35.255 ms	2 - 795 MB	NPU
DDColor	ONNX	float	Qualcomm® QCS7181	72.637 ms	150 - 150 MB	NPU
DDColor	QNN_DLC	float	Snapdragon® X2 Elite	27.882 ms	1 - 1 MB	NPU
DDColor	QNN_DLC	float	Snapdragon® X Elite	62.411 ms	1 - 1 MB	NPU
DDColor	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	41.286 ms	0 - 559 MB	NPU
DDColor	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	73.444 ms	1 - 462 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® QCS8275	129.869 ms	1 - 453 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	62.331 ms	1 - 3 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® QCS8450	73.444 ms	1 - 462 MB	NPU
DDColor	QNN_DLC	float	Snapdragon® 8 Elite Mobile	32.126 ms	1 - 617 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® SA8295P	69.553 ms	0 - 330 MB	NPU
DDColor	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	27.253 ms	0 - 643 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® SA7255P	129.869 ms	1 - 453 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® QCS9075	92.519 ms	1 - 4 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® QCS8750	32.126 ms	1 - 617 MB	NPU
DDColor	QNN_DLC	float	Qualcomm® QCS7181	62.411 ms	1 - 1 MB	NPU
DDColor	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	43.152 ms	1 - 1425 MB	NPU
DDColor	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	78.7 ms	0 - 552 MB	NPU
DDColor	TFLITE	float	Qualcomm® QCS8275	124.534 ms	1 - 799 MB	NPU
DDColor	TFLITE	float	Qualcomm® QCS8550 (Proxy)	68.491 ms	1 - 4 MB	NPU
DDColor	TFLITE	float	Qualcomm® SA8775P	463.828 ms	130 - 152 MB	CPU
DDColor	TFLITE	float	Qualcomm® SA8650P	463.828 ms	130 - 152 MB	CPU
DDColor	TFLITE	float	Qualcomm® SA8255P	463.828 ms	130 - 152 MB	CPU
DDColor	TFLITE	float	Qualcomm® QCS8450	78.7 ms	0 - 552 MB	NPU
DDColor	TFLITE	float	Snapdragon® 8 Elite Mobile	33.836 ms	1 - 983 MB	NPU
DDColor	TFLITE	float	Qualcomm® SA8295P	73.593 ms	1 - 379 MB	NPU
DDColor	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	27.463 ms	1 - 1037 MB	NPU
DDColor	TFLITE	float	Qualcomm® SA7255P	124.534 ms	1 - 799 MB	NPU
DDColor	TFLITE	float	Qualcomm® QCS9075	140.953 ms	1 - 116 MB	NPU
DDColor	TFLITE	float	Qualcomm® QCS8750	33.836 ms	1 - 983 MB	NPU

License

• The license for the original implementation of DDColor can be found here.

References

• DDColor: Towards Photo-Realistic Image Colorization via Dual Decoders
• 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.