Open lucasjinreal opened 4 years ago
From paper it claims that it supressed encoder-decoder pattern detection model, so that it can do FPN things in a single backbone, the experiements still using FPN?
From paper it claims that it supressed encoder-decoder pattern detection model, so that it can do FPN things in a single backbone, the experiements still using FPN?