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
QNN_DLC	float	Universal	QAIRT 2.45	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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
proposal_generator	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	45.237 ms	7 - 1769 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® X2 Elite	42.975 ms	7 - 7 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® X Elite	91.164 ms	7 - 7 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	69.626 ms	7 - 2329 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® QCS8275	354.065 ms	0 - 1756 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	94.839 ms	7 - 833 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA8775P	122.238 ms	2 - 1758 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA8650P	122.238 ms	2 - 1758 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA8255P	122.238 ms	2 - 1758 MB	NPU
proposal_generator	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	52.303 ms	7 - 1516 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	157.084 ms	8 - 2735 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® QCS9075	119.437 ms	7 - 71 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA8295P	125.992 ms	0 - 1430 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® SA7255P	354.065 ms	0 - 1756 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® QCS8750	52.303 ms	7 - 1516 MB	NPU
proposal_generator	QNN_DLC	float	Qualcomm® QCS7181	91.164 ms	7 - 7 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	99.499 ms	51 - 750 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® X2 Elite	97.425 ms	52 - 52 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® X Elite	242.75 ms	52 - 52 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	178.766 ms	13 - 859 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® QCS8275	582.252 ms	44 - 741 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	239.218 ms	52 - 54 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA8775P	275.574 ms	49 - 924 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA8650P	275.574 ms	49 - 924 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA8255P	275.574 ms	49 - 924 MB	NPU
roi_head	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	124.845 ms	16 - 707 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	329.843 ms	39 - 957 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® QCS9075	325.336 ms	52 - 106 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA8295P	307.262 ms	49 - 848 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® SA7255P	582.252 ms	44 - 741 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® QCS8750	124.845 ms	16 - 707 MB	NPU
roi_head	QNN_DLC	float	Qualcomm® QCS7181	242.75 ms	52 - 52 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.