Abstract: Global visual localization estimates the absolute pose of a camera using a single image, in a previously mapped area. Obtaining the pose from a single image enables many robotics and augmented/virtual reality applications. Inspired by latest advances in deep learning, many existing approaches directly learn and regress 6 DoF pose from an input image. However, these methods do not fully utilize the underlying scene geometry for pose regression. The challenge in monocular relocalization is the minimal availability of supervised training data, which is just the corresponding 6 DoF poses of the images. In this letter, we propose to utilize these minimal available labels (i.e., poses) to learn the underlying 3D geometry of the scene and use the geometry to estimate the 6 DoF camera pose. We present a learning method that uses these pose labels and rigid alignment to learn two 3D geometric representations ( X, Y, Z coordinates ) of the scene, one in camera coordinate frame and the other in global coordinate frame. Given a single image, it estimates these two 3D scene representations, which are then aligned to estimate a pose that matches the pose label. This formulation allows for the active inclusion of additional learning constraints to minimize 3D alignment errors between the two 3D scene representations, and 2D re-projection errors between the 3D global scene representation and 2D image pixels, resulting in improved localization accuracy. During inference, our model estimates the 3D scene geometry in camera and global frames and aligns them rigidly to obtain pose in real-time. We evaluate our work on three common visual localization datasets, conduct ablation studies, and show that our method exceeds state-of-the-art regression methods' pose accuracy on all datasets.

Abstract: Characterization of carotid plaque composition, more specifically the amount of lipid core, fibrous tissue, and calcified tissue, is an important task for the identification of plaques that are prone to rupture, and thus for early risk estimation of cardiovascular and cerebrovascular events. Due to its low costs and wide availability, carotid ultrasound has the potential to become the modality of choice for plaque characterization in clinical practice. However, its significant image noise, coupled with the small size of the plaques and their complex appearance, makes it difficult for automated techniques to discriminate between the different plaque constituents. In this paper, we propose to address this challenging problem by exploiting the unique capabilities of the emerging deep learning framework. More specifically, and unlike existing works which require a priori definition of specific imaging features or thresholding values, we propose to build a convolutional neural network (CNN) that will automatically extract from the images the information that is optimal for the identification of the different plaque constituents. We used approximately 90 000 patches extracted from a database of images and corresponding expert plaque characterizations to train and to validate the proposed CNN. The results of cross-validation experiments show a correlation of about 0.90 with the clinical assessment for the estimation of lipid core, fibrous cap, and calcified tissue areas, indicating the potential of deep learning for the challenging task of automatic characterization of plaque composition in carotid ultrasound.

Abstract: Depth estimation provides essential information to perform autonomous driving and driver assistance. Especially, Monocular Depth Estimation is interesting from a practical point of view, since using a single camera is cheaper than many other options and avoids the need for continuous calibration strategies as required by stereo-vision approaches. State-of-the-art methods for Monocular Depth Estimation are based on Convolutional Neural Networks (CNNs). A promising line of work consists of introducing additional semantic information about the traffic scene when training CNNs for depth estimation. In practice, this means that the depth data used for CNN training is complemented with images having pixel-wise semantic labels, which usually are difficult to annotate (eg crowded urban images). Moreover, so far it is common practice to assume that the same raw training data is associated with both types of ground truth, ie, depth and semantic labels. The main contribution of this paper is to show that this hard constraint can be circumvented, ie, that we can train CNNs for depth estimation by leveraging the depth and semantic information coming from heterogeneous datasets. In order to illustrate the benefits of our approach, we combine KITTI depth and Cityscapes semantic segmentation datasets, outperforming state-of-the-art results on Monocular Depth Estimation.

Abstract: Driver assistance and safety systems are getting attention nowadays towards automatic navigation and safety. Optical flow as a motion estimation technique has got major roll in making these systems a reality. Towards this, in the current paper, the suitability of polar representation for optical flow estimation in such systems is demonstrated. Furthermore, the influence of individual regularization terms on the accuracy of optical flow on image sequences of different speeds is empirically evaluated. Also a new synthetic dataset of image sequences with different speeds is generated along with the ground-truth optical flow.

Abstract: Common techniques for pedestrian candidates generation (e.g., sliding window approaches) are based on an exhaustive search over the image. This implies that the number of windows produced is huge, which translates into a significant time consumption in the classification stage. In this paper, we propose a method that significantly reduces the number of windows to be considered by a classifier. Our method is a monocular one that exploits geometric and depth information available on single images. Both representations of the world are fused together to generate pedestrian candidates based on an underlying model which is focused only on objects standing vertically on the ground plane and having certain height, according with their depths on the scene. We evaluate our algorithm on a challenging dataset and demonstrate its application for pedestrian detection, where a considerable reduction in the number of candidate windows is reached.

Abstract: TC8.6
This paper proposes a novel method for estimating the geospatial localization of a vehicle. I uses as input a georeferenced video sequence recorded by a forward-facing camera attached to the windscreen. The core of the proposed method is an on-line video synchronization which finds out the corresponding frame in the georeferenced video sequence to the one recorded at each time by the camera on a second drive through the same track. Once found the corresponding frame in the georeferenced video sequence, we transfer its geospatial information of this frame. The key advantages of this method are: 1) the increase of the update rate and the geospatial accuracy with regard to a standard low-cost GPS and 2) the ability to localize a vehicle even when a GPS is not available or is not reliable enough, like in certain urban areas. Experimental results for an urban environments are presented, showing an average of relative accuracy of 1.5 meters.

Abstract: Medial representations are powerful tools for describing and parameterizing the volumetric shape of anatomical structures. Accurate computation of one pixel wide medial surfaces is mandatory. Those surfaces must represent faithfully the geometry of the volume. Although morphological methods produce excellent results in 2D, their complexity and quality drops across dimensions, due to a more complex description of pixel neighborhoods. This paper introduces a continuous operator for accurate and efficient computation of medial structures of arbitrary dimension. Our experiments show its higher performance for medical imaging applications in terms of simplicity of medial structures and capability for reconstructing the anatomical volume

Abstract: Action recognition systems mostly work with videos of proper quality and resolution. Even most challenging benchmark databases for action recognition, hardly include videos of low-resolution from, e.g., surveillance cameras. In videos recorded by such cameras, due to the distance between people and cameras, people are pictured very small and hence challenge action recognition algorithms. Simple upsampling methods, like bicubic interpolation, cannot retrieve all the detailed information that can help the recognition. To deal with this problem, in this paper we combine results of bicubic interpolation with results of a state-ofthe-art deep learning-based super-resolution algorithm, through an alpha-blending approach. The experimental results obtained on down-sampled version of a large subset of Hoolywood2 benchmark database show the importance of the proposed system in increasing the recognition rate of a state-of-the-art action recognition system for handling low-resolution videos.