Divided Attention: Unsupervised Multi-Object Discovery with Contextually Separated Slots

TL;DR

This paper introduces Divided Attention (DivA), an unsupervised, real-time multi-object segmentation method that learns to separate objects in images without supervision, achieving state-of-the-art results and enabling improved classifier training.

Contribution

DivA is a novel unsupervised approach that segments multiple objects in images using a multi-modal encoder-decoder architecture with information separation, handling varying object counts and resolutions.

Findings

Achieves state-of-the-art segmentation performance.

Runs at up to 104 FPS, tripling speed of comparable methods.

Reduces performance gap from supervised methods to 12% or less.

Abstract

We investigate the emergence of objects in visual perception in the absence of any semantic annotation. The resulting model has received no supervision, does not use any pre-trained features, and yet it can segment the domain of an image into multiple independently moving regions. The resulting motion segmentation method can handle an unknown and varying number of objects in real-time. The core multi-modal conditional encoder-decoder architecture has one modality (optical flow) feed the encoder to produce a collection of latent codes (slots), and the other modality (color image) conditions the decoder to generate the first modality (flow) from the slots. The training criterion is designed to foster 'information separation' among the slots, while the architecture explicitly allocates activations to individual slots, leading to a method we call Divided Attention (DivA). At test time, DivA handles a different number of objects and different image resolution than seen at training, and is invariant to permutations of the slots. DivA achieves state-of-the-art performance while tripling the runtime speed of comparable methods, up to 104 FPS, and reduces the performance gap from supervised methods to 12% or less. Objects bootstrapped by DivA can then be used to prime static classifiers via contrastive learning. On fewer than 5,000 video clips, training DINO on DivA's object proposals narrows the performance gap to ImageNet-based training by up to 30.2% compared to training directly on the video frames.