Abstract: We present a method for the cost estimation of custom hoses from CAD models. They can come in two formats, which are easy to generate: a STL file or the image of a CAD drawing showing several orthogonal projections. The challenges in either cases are, first, to obtain from them a high level 3D description of the shape, and second, to learn a regression function for the prediction of the manufacturing time, based on geometric features of the reconstructed shape. The chosen description is the 3D line along the medial axis of the tube and the diameter of the circular sections along it. In order to extract it from STL files, we have adapted RANSAC, a robust parametric fitting algorithm. As for CAD drawing images, we propose a new technique for 3D reconstruction from data entered on any number of orthogonal projections. The regression function is a Gaussian process, which does not constrain the function to adopt any specific form and is governed by just two parameters. We assess the accuracy of the manufacturing time estimation by k-fold cross validation on 171 STL file models for which the time is provided by an expert. The results show the feasibility of the method, whereby the relative error for 80% of the testing samples is below 15%.

Abstract: During the next decade, on-board pedestrian detection systems will play a key role in the challenge of increasing traffic safety. The main target of these systems, to detect pedestrians in urban scenarios, implies overcoming difficulties like processing outdoor scenes from a mobile platform and searching for aspect-changing objects in cluttered environments. This makes such systems combine techniques in the state-of-the-art Computer Vision. In this paper we present a three module system based on both 2D and 3D cues. The first module uses 3D information to estimate the road plane parameters and thus select a coherent set of regions of interest (ROIs) to be further analyzed. The second module uses Real AdaBoost and a combined set of Haar wavelets and edge orientation histograms to classify the incoming ROIs as pedestrian or non-pedestrian. The final module loops again with the 3D cue in order to verify the classified ROIs and with the 2D in order to refine the final results. According to the results, the integration of the proposed techniques gives rise to a promising system.

Abstract: Avoiding vehicle-to-pedestrian crashes is a critical requirement for nowadays advanced driver assistant systems (ADAS) and future self-driving vehicles. Accordingly, detecting pedestrians from raw sensor data has a history of more than 15 years of research, with vision playing a central role.
During the last years, deep learning has boosted the accuracy of image-based pedestrian detectors.
However, detection is just the first step towards answering the core question, namely is the vehicle going to crash with a pedestrian provided preventive actions are not taken? Therefore, knowing as soon as possible if a detected pedestrian has the intention of crossing the road ahead of the vehicle is
essential for performing safe and comfortable maneuvers that prevent a crash. However, compared to pedestrian detection, there is relatively little literature on detecting pedestrian intentions. This paper aims to contribute along this line by presenting a new vision-based approach which analyzes the
pose of a pedestrian along several frames to determine if he or she is going to enter the road or not. We present experiments showing 750 ms of anticipation for pedestrians crossing the road, which at a typical urban driving speed of 50 km/h can provide 15 additional meters (compared to a pure pedestrian detector) for vehicle automatic reactions or to warn the driver. Moreover, in contrast with state-of-the-art methods, our approach is monocular, neither requiring stereo nor optical flow information.

Abstract: Deep video action recognition models have been highly successful in recent years but require large quantities of manually-annotated data, which are expensive and laborious to obtain. In this work, we investigate the generation of synthetic training data for video action recognition, as synthetic data have been successfully used to supervise models for a variety of other computer vision tasks. We propose an interpretable parametric generative model of human action videos that relies on procedural generation, physics models and other components of modern game engines. With this model we generate a diverse, realistic, and physically plausible dataset of human action videos, called PHAV for “Procedural Human Action Videos”. PHAV contains a total of 39,982 videos, with more than 1000 examples for each of 35 action categories. Our video generation approach is not limited to existing motion capture sequences: 14 of these 35 categories are procedurally-defined synthetic actions. In addition, each video is represented with 6 different data modalities, including RGB, optical flow and pixel-level semantic labels. These modalities are generated almost simultaneously using the Multiple Render Targets feature of modern GPUs. In order to leverage PHAV, we introduce a deep multi-task (i.e. that considers action classes from multiple datasets) representation learning architecture that is able to simultaneously learn from synthetic and real video datasets, even when their action categories differ. Our experiments on the UCF-101 and HMDB-51 benchmarks suggest that combining our large set of synthetic videos with small real-world datasets can boost recognition performance. Our approach also significantly outperforms video representations produced by fine-tuning state-of-the-art unsupervised generative models of videos.

Abstract: Designing discriminative powerful texture features robust to realistic imaging conditions is a challenging computer vision problem with many applications, including material recognition and analysis of satellite or aerial imagery. In the past, most texture description approaches were based on dense orderless statistical distribution of local features. However, most recent approaches to texture recognition and remote sensing scene classification are based on Convolutional Neural Networks (CNNs). The de facto practice when learning these CNN models is to use RGB patches as input with training performed on large amounts of labeled data (ImageNet). In this paper, we show that Local Binary Patterns (LBP) encoded CNN models, codenamed TEX-Nets, trained using mapped coded images with explicit LBP based texture information provide complementary information to the standard RGB deep models. Additionally, two deep architectures, namely early and late fusion, are investigated to combine the texture and color information. To the best of our knowledge, we are the first to investigate Binary Patterns encoded CNNs and different deep network fusion architectures for texture recognition and remote sensing scene classification. We perform comprehensive experiments on four texture recognition datasets and four remote sensing scene classification benchmarks: UC-Merced with 21 scene categories, WHU-RS19 with 19 scene classes, RSSCN7 with 7 categories and the recently introduced large scale aerial image dataset (AID) with 30 aerial scene types. We demonstrate that TEX-Nets provide complementary information to standard RGB deep model of the same network architecture. Our late fusion TEX-Net architecture always improves the overall performance compared to the standard RGB network on both recognition problems. Furthermore, our final combination leads to consistent improvement over the state-of-the-art for remote sensing scene