Lucid Data Dreaming for Video Object Segmentation

Convolutional networks reach top quality in pixel-level video object segmentation but require a large amount of training data (1k–100k) to deliver such results. We propose a new training strategy which achieves state-of-the-art results across three evaluation datasets while using 20 × – 1000 × less...

Full description

Saved in:
Bibliographic Details
Published in:International journal of computer vision 2019-09, Vol.127 (9), p.1175-1197
Main Authors: Khoreva, Anna, Benenson, Rodrigo, Ilg, Eddy, Brox, Thomas, Schiele, Bernt
Format: Article
Language:English
Subjects:
Annotations
Artificial Intelligence
Artificial neural networks
Computer Imaging
Computer Science
Domains
Dreams
Image processing
Image Processing and Computer Vision
Image segmentation
Pattern Recognition
Pattern Recognition and Graphics
Post-production processing
Stability
Training
Video data
Vision
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Convolutional networks reach top quality in pixel-level video object segmentation but require a large amount of training data (1k–100k) to deliver such results. We propose a new training strategy which achieves state-of-the-art results across three evaluation datasets while using 20 × – 1000 × less annotated data than competing methods. Our approach is suitable for both single and multiple object segmentation. Instead of using large training sets hoping to generalize across domains, we generate in-domain training data using the provided annotation on the first frame of each video to synthesize—“lucid dream” (in a lucid dream the sleeper is aware that he or she is dreaming and is sometimes able to control the course of the dream)—plausible future video frames. In-domain per-video training data allows us to train high quality appearance- and motion-based models, as well as tune the post-processing stage. This approach allows to reach competitive results even when training from only a single annotated frame, without ImageNet pre-training. Our results indicate that using a larger training set is not automatically better, and that for the video object segmentation task a smaller training set that is closer to the target domain is more effective. This changes the mindset regarding how many training samples and general “objectness” knowledge are required for the video object segmentation task.
ISSN:0920-5691
1573-1405
DOI:10.1007/s11263-019-01164-6