PoseShop: Human Image Database Construction and Personalized Content Synthesis

Abstract

We present PoseShop--a pipeline to construct segmented human image database with minimal manual intervention. By downloading, analyzing, and filtering massive amounts of human images from the Internet, we achieve a database which contains 400 thousands human figures that are segmented out of their background. The human figures are organized based on action semantic, clothes attributes, and indexed by the shape of their poses. They can be queried using either silhouette sketch or a skeleton to find a given pose. We demonstrate applications for this database for multiframe personalized content synthesis in the form of comic-strips, where the main character is the user or his/her friends. We address the two challenges of such synthesis, namely personalization and consistency over a set of frames, by introducing head swapping and clothes swapping techniques. We also demonstrate an action correlation analysis application to show the usefulness of the database for vision application.

Efficient Salient Region Detection with Soft Image Abstraction

“…This kind of decomposition is a key component of many computer vision and graphics tasks. Rather than focusing on predicting human fixation points [6,32] (another major research direction of visual attention modeling), salient region detection methods aim at uniformly highlighting entire salient object regions, thus benefiting a large number of applications, including objectof-interest image segmentation [19], adaptive compression [17], object recognition [44], content aware image editing [51], object level image manipulation [12,15,53], and internet visual media retrieval [10,11,13,29,24,23].…”

Efficient, Edge-Aware, Combined Color Quantization and Dithering

“…Exploiting this fact, visual saliency computation has been widely used in applications such as image segmentation [25], adaptive compression [26], and image retrieval [27]. Since all color quantization methods inevitably introduce approximation, it is reasonable to give priority to salient regions, while pushing the approximation error to the less important regions.…”

SalientShape: group saliency in image collections

“…All such methods compare user sketches with image edges (or boundaries), suffering from influence of background edges when finding a desired object. Salient object region extraction [11,12,34] and multiresolution region representation [32] have been used to handle background clutter. We also use explicit region information to support SBIR.…”

“…Determining characteristic or salient regions of images allows transitioning from low-level pixels to more meaningful high-level regions, and thus form an essential step for many computer graphics and computer vision applications, including interactive image editing [14,16,41,51,53], image retrieval [12,27,28,30], and internet visual media processing [11,33,34,40]. Recently, significant success has been reported in saliency-based image segmentation producing near ground-truth performance on simple images ( [1,13,15,41] and references therein).…”