Efficient Transformer Encoders for Mask2Former-style models
Paper • 2404.15244 • Published • 1
Note we would like the gating network to prioritize increasing the panoptic quality while also reducing the number of layers (to reduce the overall computations). Consequently, we introduce a utility function expressed as the linear combination of segmentation quality and the depth of the network.. Here β serves as an adaptation factor governing the trade-off between segmentation quality and computational cost... higher value of β signifies a greater emphasis on efficiency over segmentation quality.