MaskRCNN: Optimized for Qualcomm Devices

Mask R-CNN is a machine learning model that extends Faster R-CNN to perform instance segmentation by detecting objects in an image while simultaneously generating a high-quality segmentation mask for each instance. It adds a branch for predicting segmentation masks in parallel with the existing branch for bounding box recognition.

This is based on the implementation of MaskRCNN 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 2.50, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.50	Download

For more device-specific assets and performance metrics, visit MaskRCNN 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 MaskRCNN on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

• Input resolution: 800x800
• Model checkpoint: Mask R-CNN ResNet-50 FPN V2
• Model size (float): 177 MB
• Number of output classes: 91
• Number of parameters: 46.4M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
proposal_generator	ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	42.154 ms	134 - 1362 MB	NPU
proposal_generator	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	54.288 ms	116 - 1490 MB	NPU
proposal_generator	ONNX	float	Snapdragon® 8 Gen 3 Mobile	54.185 ms	143 - 2779 MB	NPU
proposal_generator	ONNX	float	Snapdragon® 8 Gen 1 Mobile	117.257 ms	129 - 2697 MB	NPU
proposal_generator	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	109.313 ms	7 - 18 MB	NPU
proposal_generator	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	76.026 ms	77 - 153 MB	NPU
proposal_generator	ONNX	float	Qualcomm® QCS8450	117.257 ms	129 - 2697 MB	NPU
proposal_generator	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	103.551 ms	7 - 18 MB	NPU
proposal_generator	ONNX	float	Qualcomm® Dragonwing™ Q-8750	54.288 ms	116 - 1490 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	42.253 ms	7 - 1447 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	55.41 ms	0 - 1619 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® X2 Elite	42.095 ms	7 - 7 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® X Elite	88.129 ms	7 - 7 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	56.523 ms	7 - 2155 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	110.376 ms	7 - 2597 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	117.484 ms	7 - 73 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	90.076 ms	7 - 11 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA8650P	122.238 ms	1 - 1758 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA8255P	122.238 ms	1 - 1758 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® QCS8450	110.376 ms	7 - 2597 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	113.877 ms	7 - 72 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	88.129 ms	7 - 7 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	55.41 ms	0 - 1619 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA8295P	114.415 ms	0 - 1302 MB	NPU
roi_head	ONNX	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	50.275 ms	199 - 565 MB	NPU
roi_head	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	62.976 ms	229 - 592 MB	NPU
roi_head	ONNX	float	Snapdragon® 8 Gen 3 Mobile	71.477 ms	238 - 684 MB	NPU
roi_head	ONNX	float	Snapdragon® 8 Gen 1 Mobile	151.524 ms	234 - 652 MB	NPU
roi_head	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	154.198 ms	108 - 208 MB	NPU
roi_head	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	91.763 ms	248 - 251 MB	NPU
roi_head	ONNX	float	Qualcomm® QCS8450	151.524 ms	234 - 652 MB	NPU
roi_head	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	112.601 ms	113 - 212 MB	NPU
roi_head	ONNX	float	Qualcomm® Dragonwing™ Q-8750	62.976 ms	229 - 592 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 For Galaxy Mobile	84.62 ms	15 - 562 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	112.735 ms	26 - 561 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® X2 Elite	90.793 ms	52 - 52 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® X Elite	225.368 ms	52 - 52 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	161.899 ms	17 - 650 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	303.324 ms	13 - 611 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	241.912 ms	52 - 107 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	227.049 ms	52 - 54 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA8650P	261.792 ms	49 - 777 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA8255P	261.792 ms	49 - 777 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® QCS8450	303.324 ms	13 - 611 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	251.797 ms	52 - 107 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	225.368 ms	52 - 52 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	112.735 ms	26 - 561 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA8295P	292.663 ms	49 - 589 MB	NPU

License

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

References

• Mask R-CNN
• 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.