Implementation Detail - Matting-Team/MattingProject GitHub Wiki

Implementation Detail

1. Our Service

Contains a description of the services we have implemented.

  • Implementation process and contents of Website that provides Matting function
  • Description of Website Design

2. Model Detail

Contains a description of the deep learning model used to provide the service.

  • Matting is performed through the synthesis of semantic features and structure features.
  • A total of three models were designed.
  1. Matting that extracts semantic information through Custom Encoder structure and adds Attention Module.
  2. Matting extracting semantic information through Pretrained MobileNetv2.