Modelpublic

qualcomm/TrOCR

sourceHugging Facemitupdated 11d agoView on Hugging Face

16likes2.9kdownloads

Model Card

TrOCR: Optimized for Qualcomm Devices

End-to-end text recognition approach with pre-trained image transformer and text transformer models for both image understanding and wordpiece-level text generation.

This is based on the implementation of TrOCR 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 1.27.1	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 [TrOCR on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/trocr).

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 TrOCR on GitHub for usage instructions.

Model Details

Model Type: Modelusecase.imagetotext

Model Stats:

—Input resolution: 320x320
—Model checkpoint: trocr-small-handwritten
—Model size (decoder) (float): 146 MB
—Model size (encoder) (float): 87.8 MB
—Number of parameters (decoder): 38.3M
—Number of parameters (encoder): 23.0M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
decoder	ONNX	float	Snapdragon® X2 Elite	1.245 ms	7 - 7 MB	NPU
decoder	ONNX	float	Snapdragon® X Elite	2.093 ms	68 - 68 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	1.471 ms	0 - 284 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Gen 1 Mobile	2.738 ms	1 - 220 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	2.796 ms	7 - 17 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	2.124 ms	1 - 3 MB	NPU
decoder	ONNX	float	Qualcomm® QCS8450	2.738 ms	1 - 220 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	2.78 ms	7 - 16 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	2.093 ms	68 - 68 MB	NPU
decoder	ONNX	float	Qualcomm® Dragonwing™ Q-8750	1.281 ms	0 - 192 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite Mobile	1.281 ms	0 - 192 MB	NPU
decoder	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1.186 ms	1 - 178 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X2 Elite	1.513 ms	7 - 7 MB	NPU
decoder	QNN_DLC	float	Snapdragon® X Elite	2.334 ms	7 - 7 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	1.369 ms	0 - 274 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	2.676 ms	2 - 218 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	2.634 ms	7 - 16 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	2.027 ms	1 - 3 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8775P	2.892 ms	5 - 106 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8650P	2.892 ms	5 - 106 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8255P	2.892 ms	5 - 106 MB	NPU
decoder	QNN_DLC	float	Qualcomm® QCS8450	2.676 ms	2 - 218 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	2.653 ms	7 - 15 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	2.334 ms	7 - 7 MB	NPU
decoder	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	1.243 ms	0 - 186 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA7255P	4.175 ms	5 - 105 MB	NPU
decoder	QNN_DLC	float	Qualcomm® SA8295P	2.618 ms	0 - 38 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Mobile	1.243 ms	0 - 186 MB	NPU
decoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.166 ms	1 - 173 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	1.378 ms	0 - 279 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	2.438 ms	0 - 218 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	2.673 ms	0 - 84 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	1.983 ms	0 - 2 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8775P	2.84 ms	0 - 104 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8650P	2.84 ms	0 - 104 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8255P	2.84 ms	0 - 104 MB	NPU
decoder	TFLITE	float	Qualcomm® QCS8450	2.438 ms	0 - 218 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	2.593 ms	0 - 83 MB	NPU
decoder	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	1.214 ms	0 - 189 MB	NPU
decoder	TFLITE	float	Qualcomm® SA7255P	4.213 ms	0 - 103 MB	NPU
decoder	TFLITE	float	Qualcomm® SA8295P	2.668 ms	0 - 44 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Mobile	1.214 ms	0 - 189 MB	NPU
decoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.143 ms	0 - 178 MB	NPU
encoder	ONNX	float	Snapdragon® X2 Elite	4.565 ms	16 - 16 MB	NPU
encoder	ONNX	float	Snapdragon® X Elite	11.285 ms	48 - 48 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Gen 3 Mobile	7.804 ms	16 - 216 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Gen 1 Mobile	19.697 ms	0 - 316 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	13.996 ms	15 - 20 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	10.875 ms	0 - 56 MB	NPU
encoder	ONNX	float	Qualcomm® QCS8450	19.697 ms	0 - 316 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	13.36 ms	15 - 19 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	11.285 ms	48 - 48 MB	NPU
encoder	ONNX	float	Qualcomm® Dragonwing™ Q-8750	5.285 ms	16 - 140 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Mobile	5.285 ms	16 - 140 MB	NPU
encoder	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	4.173 ms	2 - 127 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X2 Elite	5.171 ms	2 - 2 MB	NPU
encoder	QNN_DLC	float	Snapdragon® X Elite	12.032 ms	2 - 2 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	7.999 ms	0 - 190 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	19.543 ms	1 - 308 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	14.139 ms	2 - 12 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	11.316 ms	2 - 3 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8775P	14.024 ms	2 - 108 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8650P	14.024 ms	2 - 108 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8255P	14.024 ms	2 - 108 MB	NPU
encoder	QNN_DLC	float	Qualcomm® QCS8450	19.543 ms	1 - 308 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	13.846 ms	2 - 12 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	12.032 ms	2 - 2 MB	NPU
encoder	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	5.457 ms	2 - 116 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA7255P	37.078 ms	2 - 106 MB	NPU
encoder	QNN_DLC	float	Qualcomm® SA8295P	19.728 ms	2 - 234 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Mobile	5.457 ms	2 - 116 MB	NPU
encoder	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	4.358 ms	2 - 116 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	7.712 ms	4 - 194 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	19.536 ms	7 - 312 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	13.728 ms	7 - 67 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	10.718 ms	7 - 9 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8775P	13.479 ms	7 - 111 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8650P	13.479 ms	7 - 111 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8255P	13.479 ms	7 - 111 MB	NPU
encoder	TFLITE	float	Qualcomm® QCS8450	19.536 ms	7 - 312 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	13.342 ms	6 - 65 MB	NPU
encoder	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	5.186 ms	6 - 111 MB	NPU
encoder	TFLITE	float	Qualcomm® SA7255P	36.812 ms	7 - 119 MB	NPU
encoder	TFLITE	float	Qualcomm® SA8295P	19.064 ms	7 - 237 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Mobile	5.186 ms	6 - 111 MB	NPU
encoder	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	4.093 ms	0 - 105 MB	NPU

License

—The license for the original implementation of TrOCR can be found here.

References

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.