Unsupervised Moving Object Detection via Contextual Information Separation

• Published in: Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) pp. 879 - 888
• Main Authors: Yang, Yanchao, Loquercio, Antonio, Scaramuzza, Davide, Soatto, Stefano
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2019
• Subjects: Computer vision; Deep learning; Grouping and Shape; Mathematical models; Neural networks; Object detection; Partial differential equations; Pattern recognition; Representation Learning; Scene Analysis and Understanding; Segmentation; Statistical Learning
• DOI: 10.1109/CVPR.2019.00097

We propose an adversarial contextual model for detecting moving objects in images. A deep neural network is trained to predict the optical flow in a region using information from everywhere else but that region (context), while another network attempts to make such context as uninformative as possible. The result is a model where hypotheses naturally compete with no need for explicit regularization or hyper-parameter tuning. Although our method requires no supervision whatsoever, it outperforms several methods that are pre-trained on large annotated datasets. Our model can be thought of as a generalization of classical variational generative region-based segmentation, but in a way that avoids explicit regularization or solution of partial differential equations at run-time.