We propose an unsupervised video segmentation approach by simultaneously tracking multiple holistic figure-ground segments. Segment tracks are initialized from a pool of segment proposals generated from a figure-ground segmentation algorithm. Then, online non-local appearance models are trained incrementally for each track using a multi-output regularized least squares formulation. By using the same set of training examples for all segment tracks, a computational trick allows us to track hundreds of segment tracks efficiently, as well as perform optimal online updates in closed-form. Besides, a new composite statistical inference approach is proposed for refining the obtained segment tracks, which breaks down the initial segment proposals and recombines for better ones by utilizing high-order statistic estimates from the appearance model and enforcing temporal consistency. For evaluating the algorithm, a dataset, SegTrack v2, is collected with about 1,000 frames with pixel-level annotations. The proposed framework outperforms state-of-the-art approaches in the dataset, showing its efficiency and robustness to challenges in different video sequences.

Current Google image search adopt a hybrid search approach in which a text-based query (e.g., “Paris landmarks”) is used to retrieve a set of relevant images, which are then refined by the user (e.g., by re-ranking the retrieved images based on similarity to a selected example). We conjecture that given such hybrid image search engines, learning per-query distance functions over image features can improve the estimation of image similarity. We propose scalable solutions to learning query-specific distance functions by 1) adopting a simple large-margin learning framework, 2) using the query-logs of text-based image search engine to train distance functions used in content-based systems. We evaluate the feasibility and efficacy of our proposed system through comprehensive human evaluation, and compare the results with the state-of-the-art image distance function used by Google image search.

We address the problem of large-scale annotation of web images. Our approach is based on the concept of visual synset, which is an organization of images which are visually-similar and semantically-related. Each visual synset represents a single prototypical visual concept, and has an associated set of weighted annotations. Linear SVM's are utilized to predict the visual synset membership for unseen image examples, and a weighted voting rule is used to construct a ranked list of predicted annotations from a set of visual synsets. We demonstrate that visual synsets lead to better performance than standard methods on a new annotation database containing more than 200 million images and 300 thousand annotations, which is the largest ever reported.