Unsupervised Moving Object Detection via Contextual Information Separation
This approach provides an unsupervised solution for detecting moving objects in images, which could benefit computer vision applications by reducing reliance on annotated datasets.
The paper tackles unsupervised moving object detection by training a deep neural network to predict optical flow using contextual information, while another network makes the context uninformative, resulting in a model that outperforms supervised methods without needing explicit regularization or hyper-parameter tuning.
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.