Mask DINO: Towards A Unified Transformer-based Framework for Object Detection and Segmentation

In this paper we present Mask <PRE_TAG>DINO</POST_TAG>, a unified object detection and segmentation framework. Mask <PRE_TAG>DINO</POST_TAG> extends DINO (DETR with Improved Denoising Anchor Boxes) by adding a mask prediction branch which supports all image segmentation tasks (instance, panoptic, and semantic). It makes use of the query embeddings from DINO to dot-product a high-resolution pixel embedding map to predict a set of binary masks. Some key components in DINO are extended for segmentation through a shared architecture and training process. Mask <PRE_TAG>DINO</POST_TAG> is simple, efficient, and scalable, and it can benefit from joint large-scale detection and segmentation datasets. Our experiments show that Mask <PRE_TAG>DINO</POST_TAG> significantly outperforms all existing specialized segmentation methods, both on a ResNet-50 backbone and a pre-trained model with SwinL backbone. Notably, Mask <PRE_TAG>DINO</POST_TAG> establishes the best results to date on instance segmentation (54.5 AP on COCO), panoptic segmentation (59.4 PQ on COCO), and semantic segmentation (60.8 mIoU on ADE20K) among models under one billion parameters. Code is available at https://github.com/IDEACVR/MaskDINO.