FusionSeg: Learning to combine motion and appearance for fully automatic segmention of generic objects in videos
This addresses the need for fully automatic video object segmentation, which is incremental as it builds on existing methods by integrating motion and appearance in a unified framework.
The paper tackles the problem of segmenting generic objects in videos by learning to combine motion and appearance information, resulting in substantial improvements in state-of-the-art performance on three challenging benchmarks.
We propose an end-to-end learning framework for segmenting generic objects in videos. Our method learns to combine appearance and motion information to produce pixel level segmentation masks for all prominent objects in videos. We formulate this task as a structured prediction problem and design a two-stream fully convolutional neural network which fuses together motion and appearance in a unified framework. Since large-scale video datasets with pixel level segmentations are problematic, we show how to bootstrap weakly annotated videos together with existing image recognition datasets for training. Through experiments on three challenging video segmentation benchmarks, our method substantially improves the state-of-the-art for segmenting generic (unseen) objects. Code and pre-trained models are available on the project website.