Hugging Face's logo

qualcomm
/
MaskRCNN

Image Segmentation
PyTorch
android
Model card Files
xet
 Community
MaskRCNN / README.md
qaihm-bot's picture
qaihm-bot
v0.51.0
4cb688a verified
preview code
|
raw
history blame contribute delete
5.68 kB
metadata 
library_name: pytorch
license: other
tags:
  - android
pipeline_tag: image-segmentation

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 42.03 ms 7 - 1769 MB NPU
proposal_generator QNN_DLC float Snapdragon® X2 Elite 42.867 ms 7 - 7 MB NPU
proposal_generator QNN_DLC float Snapdragon® X Elite 91.294 ms 7 - 7 MB NPU
proposal_generator QNN_DLC float Snapdragon® 8 Gen 3 Mobile 69.641 ms 0 - 2321 MB NPU
proposal_generator QNN_DLC float Qualcomm® QCS8275 (Proxy) 353.971 ms 2 - 1759 MB NPU
proposal_generator QNN_DLC float Qualcomm® QCS8550 (Proxy) 94.892 ms 7 - 11 MB NPU
proposal_generator QNN_DLC float Qualcomm® SA8775P 122.237 ms 2 - 1758 MB NPU
proposal_generator QNN_DLC float Qualcomm® QCS9075 119.236 ms 7 - 71 MB NPU
proposal_generator QNN_DLC float Qualcomm® QCS8450 (Proxy) 149.855 ms 8 - 2737 MB NPU
proposal_generator QNN_DLC float Qualcomm® SA7255P 353.971 ms 2 - 1759 MB NPU
proposal_generator QNN_DLC float Qualcomm® SA8295P 126.005 ms 0 - 1431 MB NPU
proposal_generator QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 52.743 ms 7 - 1516 MB NPU
roi_head QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 93.53 ms 51 - 750 MB NPU
roi_head QNN_DLC float Snapdragon® X2 Elite 100.056 ms 52 - 52 MB NPU
roi_head QNN_DLC float Snapdragon® X Elite 254.926 ms 52 - 52 MB NPU
roi_head QNN_DLC float Snapdragon® 8 Gen 3 Mobile 176.4 ms 24 - 871 MB NPU
roi_head QNN_DLC float Qualcomm® QCS8275 (Proxy) 573.059 ms 39 - 735 MB NPU
roi_head QNN_DLC float Qualcomm® QCS8550 (Proxy) 243.146 ms 52 - 54 MB NPU
roi_head QNN_DLC float Qualcomm® SA8775P 270.769 ms 42 - 918 MB NPU
roi_head QNN_DLC float Qualcomm® QCS9075 321.708 ms 52 - 106 MB NPU
roi_head QNN_DLC float Qualcomm® QCS8450 (Proxy) 316.335 ms 22 - 942 MB NPU
roi_head QNN_DLC float Qualcomm® SA7255P 573.059 ms 39 - 735 MB NPU
roi_head QNN_DLC float Qualcomm® SA8295P 314.669 ms 48 - 845 MB NPU
roi_head QNN_DLC float Snapdragon® 8 Elite For Galaxy Mobile 123.558 ms 37 - 728 MB NPU

License

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

References

Community