Georg Poier

Grid Loss: Detecting Occluded Faces

Detection of partially occluded objects is a challenging computer vision problem. Standard Convolutional Neural Network (CNN) detectors fail if parts of the detection window are occluded, since not every sub-part of the window is discriminative on its own. To address this issue, we propose a novel loss layer for CNNs, named grid loss, which minimizes the error rate on sub-blocks of a convolution layer independently rather than over the whole feature map. This results in parts being more discriminative on their own, enabling the detector to recover if the detection window is partially occluded. By mapping our loss layer back to a regular fully connected layer, no additional computational cost is incurred at runtime compared to standard CNNs. We demonstrate our method for face detection on several public face detection benchmarks and show that our method outperforms regular CNNs, is suitable for realtime applications and achieves state-of-the-art performance.

Hybrid One-Shot 3D Hand Pose Estimation by Exploiting Uncertainties.

Model-based approaches to 3D hand tracking have been shown to perform well in a wide range of scenarios. However, they require initialisation and cannot recover easily from tracking failures that occur due to fast hand motions. Data-driven approaches, on the other hand, can quickly deliver a solution, but the results often suffer from lower accuracy or missing anatomical validity compared to those obtained from model-based approaches. In this work we propose a hybrid approach for hand pose estimation from a single depth image. First, a learned regressor is employed to deliver multiple initial hypotheses for the 3D position of each hand joint. Subsequently, the kinematic parameters of a 3D hand model are found by deliberately exploiting the inherent uncertainty of the inferred joint proposals. This way, the method provides anatomically valid and accurate solutions without requiring manual initialisation or suffering from track losses. Quantitative results on several standard datasets demonstrate that the proposed method outperforms state-of-the-art representatives of the model-based, data-driven and hybrid paradigms.

Interactive 3D Segmentation of Rock-Art by Enhanced Depth Maps and Gradient Preserving Regularization

Petroglyphs (rock engravings) have been pecked and engraved by humans into natural rock surfaces thousands of years ago and are among the oldest artifacts that document early human life and culture. Some of these rock engravings have survived until the present and serve today as a unique document of ancient human life. Since petroglyphs are pecked into the surface of natural rocks, they are threatened by environmental factors such as weather and erosion. To document and preserve these valuable artifacts of human history, the 3D digitization of rock surfaces has become a suitable approach due to the development of powerful 3D reconstruction techniques in recent years. The results of 3D reconstruction are huge 3D point clouds which represent the local surface geometry in high resolution. In this article, we present an automatic 3D segmentation approach that is able to extract rock engravings from reconstructed 3D surfaces. To solve this computationally complex problem, we transfer the task of segmentation to the image-space in order to efficiently perform segmentation. Adaptive learning is applied to realize interactive segmentation and a gradient preserving energy minimization assures smooth boundaries for the segmented figures. Our experiments demonstrate the efficiency and the strong segmentation capabilities of the approach. The precise segmentation of petroglyphs from 3D surfaces provides the foundation for compiling large petroglyph databases which can then be indexed and searched automatically.

Interactive segmentation of rock-art in high-resolution 3D reconstructions

Petroglyphs (rock engravings) are important artifacts for the documentation and analysis of early human life. Recent improvements in 3D scanning and 3D reconstruction enable the accurate 3D reconstruction of petroglyphs from rock surfaces at sub-millimeter resolution. To enable the indexing, matching, and recognition of petroglyphs in petroglyph databases, the shapes must first be segmented from the reconstructed rock surface. The absence of robust 3D segmentation methods for petroglyphs leaves a gap in the digital processing workflow. In this paper, we present a semi-automatic method for petroglyph segmentation for high-resolution 3D surface reconstructions. A comprehensive evaluation shows that our method is able to robustly segment petroglyphs with high accuracy and that the incorporation of 3D information is crucial to solve the segmentation problem. The presented method represents a major step towards the completion of a full 3D digital processing workflow of petroglyphs.

Multi-cue learning and visualization of unusual events

Unusual event detection, i.e., identifying unspecified rare/critical events, has become one of the major challenges in visual surveillance. The main solution for this problem is to describe local or global normalness and to report events that do not fit to the estimated models. The majority of existing approaches, however, is limited to a single description (e.g., either appearance or motion) and/or builds on inflexible (unsupervised) learning techniques, both clearly degrading the practical applicability. To overcome these limitations, we demonstrate a system that is capable of extracting and modeling several representations in parallel, while in addition allows for user interaction within a continuous learning setup. Novel yet intuitive concepts of result visualization and user interaction will be presented that allow for exploiting the underlying data.

The 3D-Pitoti Dataset: A Dataset for high-resolution 3D Surface Segmentation

The development of powerful 3D scanning hardware and reconstruction algorithms has strongly promoted the generation of 3D surface reconstructions in different domains. An area of special interest for such 3D reconstructions is the cultural heritage domain, where surface reconstructions are generated to digitally preserve historical artifacts. While reconstruction quality nowadays is sufficient in many cases, the robust analysis (e.g. segmentation, matching, and classification) of reconstructed 3D data is still an open topic. In this paper, we target the automatic segmentation of high-resolution 3D surface reconstructions of petroglyphs. To foster research in this field, we introduce a fully annotated, large-scale 3D surface dataset including high-resolution meshes, depth maps and point clouds as a novel benchmark dataset, which we make publicly available. Additionally, we provide baseline results for a random forest as well as a convolutional neural network based approach. Results show the complementary strengths and weaknesses of both approaches and point out that the provided dataset represents an open challenge for future research.

AVSS 2011 demo session: OUTLIER - online learning and visualization of unusual events

Summary form only given. We introduce to the surveillance community the VIRAT Video Dataset[1], which is a new large-scale surveillance video dataset designed to assess the performance of event recognition algorithms in realistic scenes1.

Hough Forests Revisited: An Approach to Multiple Instance Tracking from Multiple Cameras

Tracking multiple objects in parallel is a difficult task, especially if instances are interacting and occluding each other. To alleviate the arising problems multiple camera views can be taken into account, which, however, increases the computational effort. Evoking the need for very efficient methods, often rather simple approaches such as background subtraction are applied, which tend to fail for more difficult scenarios. Thus, in this work, we introduce a powerful multi-instance tracking approach building on Hough Forests. By adequately refining the time consuming building blocks, we can drastically reduce their computational complexity without a significant loss in accuracy. In fact, we show that the test time can be reduced by one to two orders of magnitude, allowing to efficiently process the large amount of image data coming from multiple cameras. Furthermore, we adapt the pre-trained generic forest model in an online manner to train an instance-specific model, making it well suited for multi-instance tracking. Our experimental evaluations show the effectiveness of the proposed efficient Hough Forests for object detection as well as for the actual task of multi-camera tracking.