VONet: Unsupervised Video Object Learning With Parallel U-Net Attention and Object-wise Sequential VAE
This addresses the problem of learning object representations from videos without labels, which is incremental as it builds on prior work like MONet.
The paper tackles unsupervised video object learning by decomposing scenes into structural object representations without supervision, achieving state-of-the-art results on five MOVI datasets with diverse complexities.
Unsupervised video object learning seeks to decompose video scenes into structural object representations without any supervision from depth, optical flow, or segmentation. We present VONet, an innovative approach that is inspired by MONet. While utilizing a U-Net architecture, VONet employs an efficient and effective parallel attention inference process, generating attention masks for all slots simultaneously. Additionally, to enhance the temporal consistency of each mask across consecutive video frames, VONet develops an object-wise sequential VAE framework. The integration of these innovative encoder-side techniques, in conjunction with an expressive transformer-based decoder, establishes VONet as the leading unsupervised method for object learning across five MOVI datasets, encompassing videos of diverse complexities. Code is available at https://github.com/hnyu/vonet.