No abstract

Abstract. In this paper we address the challenging problem of complex event recognition by using low-level events. In this problem, each complex event is captured by a long video in which several low-level events happen. The dataset contains several videos and due to the large number of videos and complexity of the events, the available annotation for the low-level events is very noisy which makes the detection task even more challenging. To tackle these problems we model the joint relationship between the low-level events in a graph where we consider a node for each low-level event and whenever there is a correlation between two low-level events the graph has an edge between the corresponding nodes. In addition, for decreasing the effect of weak and/or irrelevant low-level event detectors we consider the presence/absence of low-level events as hidden variables and learn a discriminative model by using latent SVM formulation. Using our learned model for the complex event recognition, we can also apply it for improving the detection of the low-level events in video clips which enables us to discover a conceptual description of the video. Thus our model can do complex event recognition and explain a video in terms of low-level events in a single framework. We have evaluated our proposed method over the most challenging multimedia event detection dataset. The experimental results reveals that the proposed method performs well compared to the baseline method. Further, our results of conceptual description of video shows that our model is learned quite well to handle the noisy annotation and surpass the low-level event detectors which are directly trained on the raw features.

Abstract. We present a method for multi-target tracking that exploits the persistence in detection of object parts. While the implicit representation and detection of body parts have recently been leveraged for improved human detection, ours is the first method that attempts to temporally constrain the location of human body parts with the express purpose of improving pedestrian tracking. We pose the problem of simultaneous tracking of multiple targets and their parts in a network flow optimization framework and show that parts of this network need to be optimized separately and iteratively, due to inter-dependencies of node and edge costs. Given potential detections of humans and their parts separately, an initial set of pedestrian tracklets is first obtained, followed by explicit tracking of human parts as constrained by initial human tracking. A merging step is then performed whereby we attempt to include part-only detections for which the entire human is not observable. This step employs a selective appearance model, which allows us to skip occluded parts in description of positive training samples. The result is high confidence, robust trajectories of pedestrians as well as their parts, which essentially constrain each other's locations and associations, thus improving human tracking and parts detection. We test our algorithm on multiple real datasets and show that the proposed algorithm is an improvement over the state-of-the-art.

This paper proposes direct learning of image classification from image tags in the wild, without filtering. Each wild tag is supplied by the user who shared the image online. Enormous numbers of these tags are freely available, and they give insight about the image categories important to users and to image classification. Our main contribution is an analysis of the Flickr 100 Million Image dataset, including several useful observations about the statistics of these tags. We introduce a large-scale robust classification algorithm, in order to handle the inherent noise in these tags, and a calibration procedure to better predict objective annotations. We show that freely available, wild tags can obtain similar or superior results to large databases of costly manual annotations.

Figure 1: IM2CAD takes a single photo of a real scene (left), and automatically reconstructs its 3D CAD model (right). AbstractGiven a single photo of a room and a large database of furniture CAD models, our goal is to reconstruct a scene that is as similar as possible to the scene depicted in the photograph, and composed of objects drawn from the database. We present a completely automatic system to address this IM2CAD problem that produces high quality results on challenging imagery from interior home design and remodeling websites. Our approach iteratively optimizes the placement and scale of objects in the room to best match scene renderings to the input photo, using image comparison metrics trained via deep convolutional neural nets. By operating jointly on the full scene at once, we account for inter-object occlusions. We also show the applicability of our method in standard scene understanding benchmarks where we obtain significant improvement.