In this paper, we present a robust vision-based system for vehicle tracking and classification devised traffic flow surveillance. The performs in real time, achieving good results, even challenging situations, such as with moving casted shadows on sunny days, headlight reflections the road, rainy jams, using only single standard camera. We propose adaptive multicue segmentation strategy that detects foreground pixels corresponding to stopped vehicles, noisy images due compression. First,...

A novel real-time people counting system is presented in this paper. Using a single overhead mounted camera, the counts number of going and out an observed area. Counting performed by analyzing image zone composed set virtual lines. The runs on commercial PC, does not need special background easily adjustable to different camera height requirements. We have tested performance system, achieving correct rate 95%.

Real-time depth extraction from stereo images is an important process in computer vision -- This paper proposes a new implementation of the dynamic programming algorithm to calculate dense maps using CUDA architecture achieving real-time performance with consumer graphics cards We compare running time against CPU and demonstrate scalability property by testing it on different

This article introduces a 3D vehicle tracking system in traffic surveillance environment devised for shadow tolling applications. It has been specially designed to operate real time with high correct detection and classification rates. The is capable of providing accurate robust results challenging road scenarios, rain, jams, casted shadows sunny days at sunrise sunset times, etc. A Bayesian inference method generate estimates multiple variable objects entering exiting the scene. framework...

Lane markings are mymargin a key element for Autonomous Driving. The generation of high definition maps and ground-truth data require extensive manual labor. In this paper, we present an efficient robust method the offline annotation lane markings, using low-density LIDAR point clouds odometry information. is used to accumulate scans process them blocks following trajectory vehicle. At each block, candidate marking points detected by generating virtual scan-lines applying dynamically...

Curb detection is essential for environmental awareness in Automated Driving (AD), as it typically limits drivable and non-drivable areas. Annotated data are necessary developing validating an AD function. However, the number of public datasets with annotated point cloud curbs scarce. This paper presents a method detecting 3D sequence clouds captured from LiDAR sensor, which consists two main steps. First, our approach detects at each scan using segmentation deep neural network. Then,...

In this paper we present a reliable depth estimation system which works in real-time with commodity hardware. The is specially intended for 3D visualization using autostereoscopic displays. core of work an implementation modified version the adaptive support-weight algorithm that includes highly optimized algorithms GPU, allowing accurate and stable map generation. Our approach overcomes typical problems live systems such as noise flickering. Proposed integrated within versatile GStreamer...

This paper introduces a web application for point cloud annotation that is used in the advanced driver assistance systems field. Apart from viewer, tool has an object viewer and timeline to define attributes of annotations video validate with corresponding images. The also describes several strategies we followed obtain correctly quickly: (i) memory management rendering large-scale clouds, (ii) coherent combination images annotations, (iii) content synchronization all parts (iv) automatic...

A major challenges of deep learning (DL) is the necessity to collect huge amounts training data. Often, lack a sufficiently large dataset discourages use DL in certain applications. Typically, acquiring required data costs considerable time, material and effort. To mitigate this problem, synthetic images combined with real popular approach, widely adopted scientific community effectively train various detectors. In study, we examined potential data-based field intelligent transportation...

In this paper we present a real-time 3D object tracking algorithm based on edges and using single pre-calibrated camera. During the process, is continuously projecting model to current frame by pose estimated in previous frame. Once projected, some control points are generated along visible of object. The next minimizing distances between detected image.

Keywords: adaptive isosurface extraction, tessellation, isosurfaces, volume warping, marching cubesAbstract: This work proposes a variation on the Marching Cubes algorithm, where goal is to represent implicitfunctions with higher resolution and better graphical quality using same grid size. The proposed algorithmdisplaces vertices of cubes iteratively until stop condition achieved. After each iteration, thedifference between implicit explicit representations are reduced, when algorithm...

Surface reconstruction from unorganized point set is a common problem in computer graphics -- Generation of the signed distance field methodology for surface The implicit surfaces made with algorithm marching cubes, but can not be processed cubes because unsigned nature We propose an extension to allowing 0-level iso-surfaces calculate more information inside each cell lattice and then we extract intersection points within identify case triangulation Our generates good presence ambiguities...

In this paper we investigate the use of a self-organizing map in an acoustic segmentation task. The aim is to obtain limited number classes and segment whenever change class between two adjacent frames occurs. Energy different frequency ranges used as input training process. A structure based on Kohonen connected neural network trained with back-propagation algorithm proposed.

Curb detection is essential for environmental awareness in Automated Driving (AD), as it typically limits drivable and non-drivable areas. Annotated data are necessary developing validating an AD function. However, the number of public datasets with annotated point cloud curbs scarce. This paper presents a method detecting 3D sequence clouds captured from LiDAR sensor, which consists two main steps. First, our approach detects at each scan using segmentation deep neural network. Then,...