Qualcomm® AI HubAI Hub

SINet

Lightweight portrait segmentation for background removal.

SINet is a machine learning model that is designed to segment people from close‑up portrait images in real time.

Not supported

This model is currently not supported on any IoT chipset.

To see performance metrics for this model on other chipsets, click the button below.

View for other chipsets
Model RepositoryHugging FaceResearch Paper

Technical Details

Model checkpoint:SINet.pth
Input resolution:224x224
Number of output classes:2 (foreground / background)
Number of parameters:91.9K
Model size (float):415 KB
Model size (w8a8):241 KB

Applicable Scenarios

  • Background replacement
  • Face removal

License

Model:MIT

Supported IoT Devices

  • Dragonwing IQ-9075 EVK
  • Dragonwing IQ-X5121
  • Dragonwing IQ-X7181
  • Dragonwing Q-6690 MTP
  • Dragonwing Q-7790
  • Dragonwing Q-8750
  • Dragonwing RB3 Gen 2 Vision Kit
  • QCS8275 (Proxy)
  • QCS8550 (Proxy)

Supported IoT Chipsets

  • Qualcomm® QCM6690
  • Qualcomm® QCS6490
  • Qualcomm® QCS8275 (Proxy)
  • Qualcomm® QCS8550 (Proxy)
  • Qualcomm® QCS9075

Related Models

See all models
View details for the FCN-ResNet50 model.

FCN-ResNet50

Fully-convolutional network model for image segmentation.

View details for the Unet-Segmentation model.

Unet-Segmentation

Real-time segmentation optimized for mobile and edge.

View details for the MediaPipe-Selfie-Segmentation model.

MediaPipe-Selfie-Segmentation

Segments the person from background in a selfie image and realtime background segmentation in video conferencing.

Looking for more? See models created by industry leaders.

Discover Model Makers