Coloring Action Recognition in Still Images

Abstract

In this article we investigate the problem of human action recognition in static images. By action recognition we intend a class of problems which includes both action classification and action detection (i.e. simultaneous localization and classification). Bagof-words image representations yield promising results for action classification, and deformable part models perform very well object detection. The representations for action recognition typically use only shape cues and ignore color information. Inspired by the recent success of color in image classification and object detection, we investigate the potential of color for action classification and detection in static images.We perform a comprehensive evaluation of color descriptors and fusion approaches for action recognition. Experiments were conducted on the three datasets most used for benchmarking action recognition in still images: Willow, PASCAL VOC 2010 and Stanford-40. Our experiments demonstrate that incorporating color information considerably improves recognition performance, and that a descriptor based on color names outperforms pure color descriptors. Our experiments demonstrate that late fusion of color and shape information outperforms other approaches on action recognition. Finally, we show that the different color-shape fusion approaches result in complementary information and combining them yields state-of-the-art performance for action classification.

Adaptive Color Attributes for Real-Time Visual Tracking

“…Color measurements can vary significantly over an image sequence due to variations in illuminant, shadows, shading, specularities, camera and object geometry. Robustness with respect to these factors has been studied in color imaging, and successfully applied to image classification [21,14], and action recognition [12]. Therefore, we evaluate existing color transformations for the task of visual object tracking.…”

Exemplar-Based Recognition of Human–Object Interactions

“…The human pose cue is often captured by a pictorial structural model [18] or 'poselet' [19]. To avoid unreliable pose estimation, algorithms in [20], [21], [22], [23], [24] treat action recognition as a typical image classification problem without explicitly modelling the pose of human of interest. There are also some works that combine the action recognition in still image and video together [25], [26], [27].…”

Coloring Channel Representations for Visual Tracking

“…Most stateof-the-art methods rely on solely image intensity information [17,31,8,14,35,20,7], while others employ simple color space transformations [29,27,28]. On the contrary, feature representations have been thoroughly investigated in the related fields of object recognition and action recognition [22,21]. Recently, Danelljan et al [6] introduced the Adaptive Color Tracker (ACT), which learns an adaptive color representation based on Color Names [33].…”

“…Color names have successfully been applied in object recognition [22], action recognition [21] and image stitching [26] to capture color information. On the other hand, channel coded intensity features have been used in visual tracking [31,8] to capture the image intensity statistics.…”