Associative Embedding: End-to-End Learning for Joint Detection and\n Grouping

We introduce associative embedding, a novel method for supervising\nconvolutional neural networks for the task of detection and grouping. A number\nof computer vision problems can be framed in this manner including multi-person\npose estimation, instance segmentation, and multi-object tracking. Usually the\ngrouping of detections is achieved with multi-stage pipelines, instead we\npropose an approach that teaches a network to simultaneously output detections\nand group assignments. This technique can be easily integrated into any\nstate-of-the-art network architecture that produces pixel-wise predictions. We\nshow how to apply this method to both multi-person pose estimation and instance\nsegmentation and report state-of-the-art performance for multi-person pose on\nthe MPII and MS-COCO datasets.\n