Vision Based Person Detection for Safe Navigation of Commercial Vehicle

Size: px
Start display at page:

Download "Vision Based Person Detection for Safe Navigation of Commercial Vehicle"

Transcription

1 Vision Based Person Detection for Safe Navigation of Commercial Vehicle Songlin Piao and Karsten Berns University of Kaiserslautern, 67663, Germany, Abstract. Vision based solution of safe navigation for commercial vehicles with fish-eye camera is presented in this paper. This work aims to develop a system which is able to detect persons or objects around commercial vehicles for preventing any accidents. This is achieved by integrating classifier based window searching algorithm and ego-motion based algorithm into the system. The classifier is trained using cascaded support vector machine and ego-motion is estimated only based on captured images from camera. Test results show that designed system provides high detection rate when it is applied in commercial vehicle. 1 Introduction Nowadays, safety issue is becoming more and more important in autonomous mobile robots. The word Autonomous receives increasing popularities in the industry of mobile vehicles. One of the main challenges ahead of industry is to promote the safety rate during the autonomous operation. Vehicle manufactures try to reduce accident rate by integrating a diverse range of safety techniques into their product. However, it seems not easy for commercial vehicles to fulfill this requirement. There were 4216 fatalities caused by large trucks and buses in U.S. in 2012 according to Motor Carrier Safety Progress Report [1]. The reason of such kind of accidents can be categorized into two main types. First, there are many blind areas which driver cannot observe directly around these machines because of limited vision scope, the tremendous size and diverse shapes. Second, though various sensors are mounted around the vehicle to detect blind spots, the sensory systems posses only the functionality of measuring data and are not able to process them meanwhile. These realistic problems motivated us to develop and design safety system described in those scenarios. Similar as other object detection tasks, human detection is a challenging endeavor because of Sensing noise, Environmental variations, Similarity to background signal, Appearance variability and unpredictability, Similarity to other people and Active deception according to the statement [2]. Environmental variations may be the most significant factor due to unexpected or abrupt changes in outdoor environment affecting directly the quality of data sensing. For example, sudden variations of lighting, foggy, snowing or raining weather conditions could

2 2 introduce tremendous noises to the sensory system especially to vision systems. These kind of factors would directly have a strong impact on higher false positive rate and lower detection rate. False positive rate decides the usability whereas the detection rate leads to the performance of the safety system. Various types of sensors have been used to detect humans so far. Radar sensors are widely utilized in the automotive industry nowadays since it was first successfully implemented in the military industry decades ago. Typical long range radar sensor from Bosch is now already in 4th generation since its first introduction into the market in 1990 s [3]. Adaptive Cruise Control which is available in most cars nowadays employs this sensor. The radar sensors are also applied in human detection in [4], [5]. There are also system implementations which used laser scanners and thermal infrared systems as alternative ones. Braun et al. detected human face with RGB camera and tracked human legs using laser scanner in [6], and Fuerstenberg developed pedestrian protection system with only laser scanner in [7]. Although all of these sensors could achieve relatively high detection rate, they lose most of visible information so that it is not suitable for short range object detection. Instead, vision sensors capture all visible data and cost relatively cheaper. Based on real time requirements, one of the main problems is the reduction of computational effect of the detection algorithm. We managed to solve this problem by applying linear support vector machine (SVM) and integral images. Discussing about detection algorithms, a classifier needs to be trained offline for the specific object. Two kinds of greedy searching methods are mainly used. One is to fix the size of searching window and to change the size of the whole image in each pyramid layer. The other one is to fix the size of the whole image and to adapt the size of searching window for each scanning level. Dalal and Triggs used histogram of oriented gradients (HOG) to detect human [8]. They trained human detection classifier using linear SVM and used first type of searching method to scan whole image. Instead, Viola and Jones applied Haar-like features and the second searching strategy mentioned above to scan face area in the image [9]. The classifier was trained using Adaboost and the program could run in real time. The trick they used is integral image which could calculate sum of the pixels inside specific region within constant time complexity. The performance of the detection algorithm depends on features used to extract descriptor, length of descriptor, learning algorithm and window scanning strategy. The remainder of this paper is organized as follows. Section 2 introduces fisheye camera and its mathematical model used in the system. Section 3 discusses local binary pattern (LBP) feature applied into the system and window scanning algorithm. Section 4 introduces training procedure of the linear and kernel intersection SVM and its potential problems more in detail. Section 5 will show the process of estimating ego-motion of a vehicle relative to the ground plane based on vision sensor and the principles of detecting foreground object based on this ego-motion. The experimental result will be described in section 6 and section 7 gives conclusion and future work.

3 3 (a) Mobotix Camera (b) Captured Image (c) Re-projection Error (d) Calibration Fig. 1. Vision System Overview and Calibration Results 2 Fish-Eye Vision Sensor One of the big advantages of fish-eye camera is the view range of 360 degree. Comparing with stereo cameras which need too much time on calculating depth image in preprocessing step, only one calibration step is enough in fish-eye camera. Once it is calibrated and the height of camera is known, then the distance between detected object and the point which corresponds to the center of camera could be estimated. There are only a few literatures available on fish-eye camera. Scaramuzza et al. proposed omni-directional visual odometry in [10], and Gandhi and Trivedi used omni-camera to detect interesting events such as independently moving persons and vehicles in [11] by estimating parametric planar motion model. Both of them are not focused on solving safety issues using such sensor.

4 4 The vision sensor used in this work is Mobotix Hemispheric Camera 1 as shown in Fig. 1(a). It can provide 180 degree horizontal field of view and 160 degree vertical field of view. One of the troublesome problems is that it is difficult to demonstrate human in the posture of upright in the fish-eye image scene, but most of state-of-art algorithms assume that humans are upright in the image scene. Therefore, it is necessary to do a pre-processing step to get a normal view. We exploit the algorithm and toolbox from [12] to calibrate the camera to get mapping relationship between world coordinate and camera coordinate. Fig. 1(b) is one of the image used for calibration with chessboard; Fig. 1(c) shows the distribution of re-projection error of each point for all the chessboard, different colors correspond to the different images of the chessboard; Fig. 1(d) displays the position of every chessboard with respect to the reference frame of the fish-eye camera. Then the critical area predefined inside omni-view can be mapped to the normal view with these parameters. In this procedure the reverse mapping technique is chosen to overcome aliasing effect. Final image processing will be done on the image size of 720 by 240 as shown in the right-handed side of Fig. 2. Width of the right image corresponds to angle inside left one and height of the right image corresponds to radius inside left one. It is not necessary to map all omni-view image but interesting area into the perspective view because the distortion is very large on the pixels which are near to the center of omni view. In the case of large distortion, sensors like ultrasonic or other vision based algorithms can be applied to detect objects in such area. Each region of interest could be represented by four parameters like angle min, angle max, radius min and radius max. The unit of angle and radius are respectively radian and pixel. These values are set to radian, radian, 40.0pixel, 270.0pixel in the designed system. Fig. 2. Projecting the image in fish-eye camera to a landscape view 3 Classifier based Detection A lot of work about vision based human detection have been done by other researchers. In the past, template matching based methods were used to detect 1

5 5 (a) (b) Fig. 3. Local Binary Pattern Descriptor humans inside the image, but recently, descriptor based methods are becoming preferring choice for human detection. A human detection classifier is trained by learning extracted positive and negative descriptors from the samples of training dataset. Then greedy searching strategy is applied to scan whole image with pre-trained classifier to judge the existence of human in specific position. Since HOG [8] was first proposed in 2005, many descriptors were proposed to get optimized balance between speed and accuracy. Some of the researchers tried to use GPU to get higher speed, but these kinds of implementations were computationally expensive so that it is not suitable for on-board robotic system. Wu et al. proposed a real-time human detection algorithm using contour cues with CENTRIST descriptor [13]. Inspired by [13], contour information is further used in the designed system to extract human descriptor, but Local Binary pattern (LBP) replaces the original CENTRIST feature. As seen in Fig. 3, each patch is divided into several grids, and then green box scans the whole patch 2 and makes final query descriptor. In case of Fig. 3, the patch is divided to 3 by 4 small grids. For each green box, the length of LBP descriptor is 256 and the total length of the final descriptor is = 1536 as there are 2 3 = 6 possible positions of green box. For each pixel inside green box the index of LBP descriptor can be calculated as shown in Fig. 3(b). The index can be represented as (C 1 C 2 C 3 C 4 C 5 C 6 C 7 C 8 ) 2 where C i is set to 1 if the corresponding neighbor pixel value is higher than the current pixel value otherwise it is set to 0. Cascaded detector which combines linear SVM and histogram intersection kernel SVM(HIK SVM) is trained at the end as shown in Fig. 4. Human detector which has patch size of 90 by 40 and 108 by 36 have been trained, respectively. 2 Here patch is Sobel edge image.

6 6 Fig. 4. Cascade in Detection 4 Training Procedure SVM is proven as a good option for binary classification problem. As discussed in the previous section, there are two types of SVM, which are applied in the system; One is linear SVM and the other is HIK SVM. Given a training set which consists of positive set P and negative set N, linear SVM is trained by iterative approach. Each sample in the training set is represented by the extracted feature vector described in the previous section. In the first iteration, the classifier is trained by P and randomly selected subset N 1 from N. We note this classifier as L 1. Similarly, L 2 is trained by P and N 2, where N 2 is the hard example on the negative set N produced by classifier L 1. In general, classifier L n is trained by P and N n, where N n is the collection of the hard example which is classified as positive by all the classifiers L 1,, L n 1. Then the final linear SVM could be trained by the positive sample set P and combination of negative sample set from each iteration procedure, which can be denoted as N final = N 1,, N n. However, there is some problem in this linear SVM. If this final trained classifier L final would be applied to other negative test dataset N test, it is of high possibility to have hard examples inside the indistinct area as shown in Fig. 5(b). This indistinct area could be depicted as two positive and negative support vectors as shown in Fig. 5(a). HIK SVM is designed to solve this problem, it is trained by positive set P and hard examples produced on the new negative test set N test by final linear SVM classifier L final. By cascading these two types of SVM, false positive rate could be reduced to much lower level and precision of the classifier also remains as high as possible, but this cascaded structure cannot affect recall rate instead. According to our research, recall rate mainly depends on the selection of positive training set P. More details about recall rate would be discussed in the experiment section.

7 7 (a) (b) Fig. 5. SVM Training 5 Ego-Motion based Detection Ego-motion is very important for the vehicle localization, especially in absence of GPS information or with slowly updating frequency. This algorithm is used to detect objects with independent motion or with obviously different height to the ground plane. Gandhi and Trivedi proposed similar method in [11], but they considered parametric planer motion model, which means they first did transformation from omni-view to perspective view and then estimated the homography. On the contrary, homography matrix is estimated directly in the proposed system, which is faster than the previous one. If X 1 and X 2 are coordinates of the same scene point in camera frame 1 and 2, then their relationship can be represented as X 2 = RX 1 + T (1) where R and T are rotation and translation matrix. If assume X 1 lies on the plane represented by normal vector n R 3, then the relationship can be written as n T X 1 = h, (2) substituting Eq. (2) into Eq. (1), we get X 2 = (R + T nt h )X 1. (3) Eq. (3) is the definition of homogeneous relationship between X 1 and X 2 and this can be simply written as X 2 = HX 1. (4) It is important to mention that in Eq. (4) scaling factor is already considered. First, good corner points are calculated by Shi-Tomasi corner detection

8 8 (a) (b) Fig. 6. Ego Motion Example algorithm [14], and then these points are tracked by optical flow based KLT feature tracker. The result of this first part is shown in Fig. 6(a). Random sample consensus (RANSAC) algorithm is applied to estimate accurate homography matrix in the next step for filtering out possible errors. After estimation of final homography matrix H, back projection needs to be implemented from current frame to the previous frame using X1 = H 1 X2 to calculate corresponding point in the previous frame. Using Eq. (5), region which has large difference with current pixel value can be estimated. The result is shown in Fig. 6(b). Red points correspond to the region which may contain foreground. f (X2 ) f (X1 ) > θ, X1 F rameprevious (5) Post processing is needed after this step. First, eroding and dilating operations are applied in order to filter out too small errors. Second, contour searching algorithm is further applied on the result image from first step to extract possible shape information of the object. Third, several conditions are further checked for each detected contour. These conditions are summarized as below: If the size of the contour is too small or too big, then it is filtered out. If the ratio between area of contour and area of the minimum enclosing circle of that contour is too small, then the contour is filtered out. Fig. 7 shows the result of ego-motion based algorithm. Fig. 7(a) shows the analyzed result of optical flow; blue points are the outliers calculated by the RANSAC algorithm. Most of the outliers are on the boundary of the image and vehicle itself. Fig. 7(b) shows the middle result after back projection of homography matrix H. There are a lot of red points on the boundary of the vehicle because of the vibration. Fig. 7(c) shows the final result where possible contour is colored as white and false contour is colored as gray. The test video used in Fig. 7 consists of 198 frames, where there are in total 76 people appeared around the machine. Our Ego Motion algorithm successfully detects humans 65 times with 2 times false detection, which means the precision

9 9 (a) Optical Flow (b) Foreground (c) Final Result Fig. 7. Ego Motion Result and recall rate of the proposed algorithm on this video are about 97% and 86%, respectively. 6 EXPERIMENTS Currently, there is no public benchmark dataset available for fish-eye camera, therefore, we have made ground truth by ourselves. There are in total 430 frames in this test sequence. Test results from designed human detection algorithm and ego-motion based algorithm are shown in Table 1. Area (Rq Rg ) >0.5 Area (Rq Rg ) (6) Matching criteria from Eq. (6) is used to compare two rectangles for the detection algorithm. The overall recall rate of the detection algorithm is 55.47%, but it becomes 70.46% when the distance between human and the camera is in the range of 1 3m because of minimal distortion in this range in the transformed image. Here the distance describes the journey between object and the point which go through the center point of camera vertically.

10 10 Fig. 8 shows the recall rate of human detection algorithm in different ranges more in detail. The ranges used in the test are 0 1m, 1 2m, 2 3m, 3 4m, 4 5m, 5 8m. Normally, there is large distortion in the transformed image in the near distance range; on the other point, if distance is too far away from camera, then objects are observed too small to be detected. This is why the recall rates in the range from 0m to 1m and 5m to 8m are very low. The recall rate inside range from 3m to 4m is relatively lower than other ranges of before and after because there were two humans appeared in this range, where occluded human was counted as a missing object in this case. Ego-motion based detector is designed in order to increase the recall rate within the range where distortion is very large so that system could still detect humans while normal human detection algorithm would not work well. The recall rate of Ego-motion based detector reaches 91.38% in the test set. The correctness of this algorithm is based on the estimated homogeneous matrix H as already shown in Eq. 4. In this test set, there were totally 429 homogeneous matrix H computed where 37 of them were falsely estimated. This happens because of abrupt change of sunlight, shadow of the vehicle and vibration of the vehicle. Detection (Overall) Detection (1m - 3m) Ego Motion Recall 55.47% 70.46% 91.38% False Positive Per Frame Average Time Per Frame ms ms ms Table 1. Test Results Fig. 9 is kind of visual result of designed system. After detection of humans, distance from vehicle and height of each detected human will be estimated using camera calibration information as in Fig.9(a) and then the results were drawn in the original image as illustrated in Fig. 9(b). Fig. 9(c) is showing the radar information of which objects around vehicle can be perceived. 7 CONCLUSIONS AND FUTURE WORK A safety system for commercial vehicle using fish-eye camera is proposed. The significant contribuitons of this paper can be summerized as follows. This is the first research that employs fish-eye camera to solve the safety issues for the commercial vehicles. Classifier and ego-motion based detection apporaches are respectively introduced, implemented and compared as well inbetween based on experimental results. In order to promote the detection rate on commercial vehicles, a fusion of both methods could be the next target in the scope as well as the object learning from different viewing angles.

11 11 Fig. 8. Recall Rate Analysis In Different Range Acknowledgements This research has been supported by European Union Seventh Framework Program (FP7/ ) under grant agreement number and German Federal Ministry of Education and Research (BMBF). References 1. Administration, F.M.C.S.: Motor carrier safety progress report (as of september 30, 2013). Technical report 2. Teixeira, T., Dublon, G., Savvides, A.: A survey of human-sensing: Methods for detecting presence, count, location, track, and identity. (2010) 3. Bosch: Fourth-generation long-range radar 4. Chang, S., Mitsumoto, N., Burdick, J.: An algorithm for uwb radar-based human detection. In: Radar Conference, 2009 IEEE. (May 2009) Tutusaus, M., Koponen, S.: Evaluation of Automotive Commercial Radar for Human Detection. (2008) 6. Braun, T., Szentpetery, K., Berns, K.: Detecting and following humans with a mobile robot. In: Proceedings of the EOS Conference On Industrial Imaging and Machine Vision. (June 2005) 7. Fuerstenberg, K.: Pedestrian protection using laserscanners. In: Intelligent Transportation Systems, Proceedings IEEE. (Sept 2005) Dalal, N., Triggs, B.: Histograms of oriented gradients for human detection. In Schmid, C., Soatto, S., Tomasi, C., eds.: International Conference on Computer Vision & Pattern Recognition. Volume 2., INRIA Rhône-Alpes, ZIRST-655, av. de l Europe, Montbonnot (June 2005) Viola, P., Jones, M.J.: Robust real-time face detection. Int. J. Comput. Vision 57(2) (May 2004)

12 12 (a) (b) (c) Fig. 9. Result Show 10. Scaramuzza, D., Siegwart, R.: Appearance-guided monocular omnidirectional visual odometry for outdoor ground vehicles. Robotics, IEEE Transactions on 24(5) (2008) Gandhi, T., Trivedi, M.: Parametric ego-motion estimation for vehicle surround analysis using an omnidirectional camera. Machine Vision and Applications 16 (2005) Scaramuzza, D., Martinelli, A., Siegwart, R.: A flexible technique for accurate omnidirectional camera calibration and structure from motion. In: Proceedings of the Fourth IEEE International Conference on Computer Vision Systems. ICVS 06, Washington, DC, USA, IEEE Computer Society (2006) Wu, J., Tan, W.C., Rehg, J.M.: Efficient and effective visual codebook generation using additive kernels. J. Mach. Learn. Res (November 2011) Shi, J., Tomasi, C.: Good features to track. In: 9th IEEE Conference on Computer Vision and Pattern Recognition, Springer-Verlag (1994)

Multi-Object Tracking Based on Tracking-Learning-Detection Framework

Multi-Object Tracking Based on Tracking-Learning-Detection Framework Multi-Object Tracking Based on Tracking-Learning-Detection Framework Songlin Piao, Karsten Berns Robotics Research Lab University of Kaiserslautern Abstract. This paper shows the framework of robust long-term

More information

Human Detection. A state-of-the-art survey. Mohammad Dorgham. University of Hamburg

Human Detection. A state-of-the-art survey. Mohammad Dorgham. University of Hamburg Human Detection A state-of-the-art survey Mohammad Dorgham University of Hamburg Presentation outline Motivation Applications Overview of approaches (categorized) Approaches details References Motivation

More information

Mobile Human Detection Systems based on Sliding Windows Approach-A Review

Mobile Human Detection Systems based on Sliding Windows Approach-A Review Mobile Human Detection Systems based on Sliding Windows Approach-A Review Seminar: Mobile Human detection systems Njieutcheu Tassi cedrique Rovile Department of Computer Engineering University of Heidelberg

More information

Large-Scale Traffic Sign Recognition based on Local Features and Color Segmentation

Large-Scale Traffic Sign Recognition based on Local Features and Color Segmentation Large-Scale Traffic Sign Recognition based on Local Features and Color Segmentation M. Blauth, E. Kraft, F. Hirschenberger, M. Böhm Fraunhofer Institute for Industrial Mathematics, Fraunhofer-Platz 1,

More information

Object detection using non-redundant local Binary Patterns

Object detection using non-redundant local Binary Patterns University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 2010 Object detection using non-redundant local Binary Patterns Duc Thanh

More information

Computer Vision with MATLAB MATLAB Expo 2012 Steve Kuznicki

Computer Vision with MATLAB MATLAB Expo 2012 Steve Kuznicki Computer Vision with MATLAB MATLAB Expo 2012 Steve Kuznicki 2011 The MathWorks, Inc. 1 Today s Topics Introduction Computer Vision Feature-based registration Automatic image registration Object recognition/rotation

More information

Person Detection in Images using HoG + Gentleboost. Rahul Rajan June 1st July 15th CMU Q Robotics Lab

Person Detection in Images using HoG + Gentleboost. Rahul Rajan June 1st July 15th CMU Q Robotics Lab Person Detection in Images using HoG + Gentleboost Rahul Rajan June 1st July 15th CMU Q Robotics Lab 1 Introduction One of the goals of computer vision Object class detection car, animal, humans Human

More information

Human Motion Detection and Tracking for Video Surveillance

Human Motion Detection and Tracking for Video Surveillance Human Motion Detection and Tracking for Video Surveillance Prithviraj Banerjee and Somnath Sengupta Department of Electronics and Electrical Communication Engineering Indian Institute of Technology, Kharagpur,

More information

FAST HUMAN DETECTION USING TEMPLATE MATCHING FOR GRADIENT IMAGES AND ASC DESCRIPTORS BASED ON SUBTRACTION STEREO

FAST HUMAN DETECTION USING TEMPLATE MATCHING FOR GRADIENT IMAGES AND ASC DESCRIPTORS BASED ON SUBTRACTION STEREO FAST HUMAN DETECTION USING TEMPLATE MATCHING FOR GRADIENT IMAGES AND ASC DESCRIPTORS BASED ON SUBTRACTION STEREO Makoto Arie, Masatoshi Shibata, Kenji Terabayashi, Alessandro Moro and Kazunori Umeda Course

More information

Seminar Heidelberg University

Seminar Heidelberg University Seminar Heidelberg University Mobile Human Detection Systems Pedestrian Detection by Stereo Vision on Mobile Robots Philip Mayer Matrikelnummer: 3300646 Motivation Fig.1: Pedestrians Within Bounding Box

More information

PEOPLE IN SEATS COUNTING VIA SEAT DETECTION FOR MEETING SURVEILLANCE

PEOPLE IN SEATS COUNTING VIA SEAT DETECTION FOR MEETING SURVEILLANCE PEOPLE IN SEATS COUNTING VIA SEAT DETECTION FOR MEETING SURVEILLANCE Hongyu Liang, Jinchen Wu, and Kaiqi Huang National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Science

More information

A New Strategy of Pedestrian Detection Based on Pseudo- Wavelet Transform and SVM

A New Strategy of Pedestrian Detection Based on Pseudo- Wavelet Transform and SVM A New Strategy of Pedestrian Detection Based on Pseudo- Wavelet Transform and SVM M.Ranjbarikoohi, M.Menhaj and M.Sarikhani Abstract: Pedestrian detection has great importance in automotive vision systems

More information

Object Detection Design challenges

Object Detection Design challenges Object Detection Design challenges How to efficiently search for likely objects Even simple models require searching hundreds of thousands of positions and scales Feature design and scoring How should

More information

Chapter 9 Object Tracking an Overview

Chapter 9 Object Tracking an Overview Chapter 9 Object Tracking an Overview The output of the background subtraction algorithm, described in the previous chapter, is a classification (segmentation) of pixels into foreground pixels (those belonging

More information

Computer Science Faculty, Bandar Lampung University, Bandar Lampung, Indonesia

Computer Science Faculty, Bandar Lampung University, Bandar Lampung, Indonesia Application Object Detection Using Histogram of Oriented Gradient For Artificial Intelegence System Module of Nao Robot (Control System Laboratory (LSKK) Bandung Institute of Technology) A K Saputra 1.,

More information

Real-Time Human Detection using Relational Depth Similarity Features

Real-Time Human Detection using Relational Depth Similarity Features Real-Time Human Detection using Relational Depth Similarity Features Sho Ikemura, Hironobu Fujiyoshi Dept. of Computer Science, Chubu University. Matsumoto 1200, Kasugai, Aichi, 487-8501 Japan. si@vision.cs.chubu.ac.jp,

More information

Human detection solution for a retail store environment

Human detection solution for a retail store environment FACULDADE DE ENGENHARIA DA UNIVERSIDADE DO PORTO Human detection solution for a retail store environment Vítor Araújo PREPARATION OF THE MSC DISSERTATION Mestrado Integrado em Engenharia Eletrotécnica

More information

Detection of a Single Hand Shape in the Foreground of Still Images

Detection of a Single Hand Shape in the Foreground of Still Images CS229 Project Final Report Detection of a Single Hand Shape in the Foreground of Still Images Toan Tran (dtoan@stanford.edu) 1. Introduction This paper is about an image detection system that can detect

More information

Calibration of a Different Field-of-view Stereo Camera System using an Embedded Checkerboard Pattern

Calibration of a Different Field-of-view Stereo Camera System using an Embedded Checkerboard Pattern Calibration of a Different Field-of-view Stereo Camera System using an Embedded Checkerboard Pattern Pathum Rathnayaka, Seung-Hae Baek and Soon-Yong Park School of Computer Science and Engineering, Kyungpook

More information

Laserscanner Based Cooperative Pre-Data-Fusion

Laserscanner Based Cooperative Pre-Data-Fusion Laserscanner Based Cooperative Pre-Data-Fusion 63 Laserscanner Based Cooperative Pre-Data-Fusion F. Ahlers, Ch. Stimming, Ibeo Automobile Sensor GmbH Abstract The Cooperative Pre-Data-Fusion is a novel

More information

Fast Human Detection for Indoor Mobile Robots Using Depth Images

Fast Human Detection for Indoor Mobile Robots Using Depth Images 2013 IEEE International Conference on Robotics and Automation (ICRA) Karlsruhe, Germany, May 6-10, 2013 Fast Human Detection for Indoor Mobile Robots Using Depth Images Benjamin Choi 1, Çetin Meriçli 1,

More information

Implementation of a Pedestrian Detection Device based on CENTRIST for an Embedded Environment

Implementation of a Pedestrian Detection Device based on CENTRIST for an Embedded Environment , pp.123-127 http://dx.doi.org/10.14257/astl.2014.46.29 Implementation of a Pedestrian Detection Device based on CENTRIST for an Embedded Environment Yun-Seop Hwang 1, Jae-Chang Kwak 2, Kwang-Yeob Lee

More information

arxiv: v1 [cs.cv] 27 Jan 2015

arxiv: v1 [cs.cv] 27 Jan 2015 A Cheap System for Vehicle Speed Detection Chaim Ginzburg, Amit Raphael and Daphna Weinshall School of Computer Science and Engineering, Hebrew University of Jerusalem, Israel arxiv:1501.06751v1 [cs.cv]

More information

Co-occurrence Histograms of Oriented Gradients for Pedestrian Detection

Co-occurrence Histograms of Oriented Gradients for Pedestrian Detection Co-occurrence Histograms of Oriented Gradients for Pedestrian Detection Tomoki Watanabe, Satoshi Ito, and Kentaro Yokoi Corporate Research and Development Center, TOSHIBA Corporation, 1, Komukai-Toshiba-cho,

More information

A novel template matching method for human detection

A novel template matching method for human detection University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 2009 A novel template matching method for human detection Duc Thanh Nguyen

More information

Bus Detection and recognition for visually impaired people

Bus Detection and recognition for visually impaired people Bus Detection and recognition for visually impaired people Hangrong Pan, Chucai Yi, and Yingli Tian The City College of New York The Graduate Center The City University of New York MAP4VIP Outline Motivation

More information

Human detection using local shape and nonredundant

Human detection using local shape and nonredundant University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 2010 Human detection using local shape and nonredundant binary patterns

More information

CS 231A Computer Vision (Winter 2014) Problem Set 3

CS 231A Computer Vision (Winter 2014) Problem Set 3 CS 231A Computer Vision (Winter 2014) Problem Set 3 Due: Feb. 18 th, 2015 (11:59pm) 1 Single Object Recognition Via SIFT (45 points) In his 2004 SIFT paper, David Lowe demonstrates impressive object recognition

More information

Face Tracking in Video

Face Tracking in Video Face Tracking in Video Hamidreza Khazaei and Pegah Tootoonchi Afshar Stanford University 350 Serra Mall Stanford, CA 94305, USA I. INTRODUCTION Object tracking is a hot area of research, and has many practical

More information

CS 231A Computer Vision (Winter 2018) Problem Set 3

CS 231A Computer Vision (Winter 2018) Problem Set 3 CS 231A Computer Vision (Winter 2018) Problem Set 3 Due: Feb 28, 2018 (11:59pm) 1 Space Carving (25 points) Dense 3D reconstruction is a difficult problem, as tackling it from the Structure from Motion

More information

Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers

Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers A. Salhi, B. Minaoui, M. Fakir, H. Chakib, H. Grimech Faculty of science and Technology Sultan Moulay Slimane

More information

Category vs. instance recognition

Category vs. instance recognition Category vs. instance recognition Category: Find all the people Find all the buildings Often within a single image Often sliding window Instance: Is this face James? Find this specific famous building

More information

Classification of objects from Video Data (Group 30)

Classification of objects from Video Data (Group 30) Classification of objects from Video Data (Group 30) Sheallika Singh 12665 Vibhuti Mahajan 12792 Aahitagni Mukherjee 12001 M Arvind 12385 1 Motivation Video surveillance has been employed for a long time

More information

3D object recognition used by team robotto

3D object recognition used by team robotto 3D object recognition used by team robotto Workshop Juliane Hoebel February 1, 2016 Faculty of Computer Science, Otto-von-Guericke University Magdeburg Content 1. Introduction 2. Depth sensor 3. 3D object

More information

Vehicle Tracking using Optical Features

Vehicle Tracking using Optical Features Vehicle Tracking using Optical Features Songlin Piao Robotics Research Lab University of Kaiserslautern, Germany Outline Motivation State-of-Art Proposed System Experiments Conclusion Motivation Autonomous

More information

Real Time Stereo Vision Based Pedestrian Detection Using Full Body Contours

Real Time Stereo Vision Based Pedestrian Detection Using Full Body Contours Real Time Stereo Vision Based Pedestrian Detection Using Full Body Contours Ion Giosan, Sergiu Nedevschi, Silviu Bota Technical University of Cluj-Napoca {Ion.Giosan, Sergiu.Nedevschi, Silviu.Bota}@cs.utcluj.ro

More information

Fast and Stable Human Detection Using Multiple Classifiers Based on Subtraction Stereo with HOG Features

Fast and Stable Human Detection Using Multiple Classifiers Based on Subtraction Stereo with HOG Features 2011 IEEE International Conference on Robotics and Automation Shanghai International Conference Center May 9-13, 2011, Shanghai, China Fast and Stable Human Detection Using Multiple Classifiers Based on

More information

Object Category Detection: Sliding Windows

Object Category Detection: Sliding Windows 04/10/12 Object Category Detection: Sliding Windows Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem Today s class: Object Category Detection Overview of object category detection Statistical

More information

Fast Human Detection Algorithm Based on Subtraction Stereo for Generic Environment

Fast Human Detection Algorithm Based on Subtraction Stereo for Generic Environment Fast Human Detection Algorithm Based on Subtraction Stereo for Generic Environment Alessandro Moro, Makoto Arie, Kenji Terabayashi and Kazunori Umeda University of Trieste, Italy / CREST, JST Chuo University,

More information

https://en.wikipedia.org/wiki/the_dress Recap: Viola-Jones sliding window detector Fast detection through two mechanisms Quickly eliminate unlikely windows Use features that are fast to compute Viola

More information

Using temporal seeding to constrain the disparity search range in stereo matching

Using temporal seeding to constrain the disparity search range in stereo matching Using temporal seeding to constrain the disparity search range in stereo matching Thulani Ndhlovu Mobile Intelligent Autonomous Systems CSIR South Africa Email: tndhlovu@csir.co.za Fred Nicolls Department

More information

A Hybrid Face Detection System using combination of Appearance-based and Feature-based methods

A Hybrid Face Detection System using combination of Appearance-based and Feature-based methods IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.5, May 2009 181 A Hybrid Face Detection System using combination of Appearance-based and Feature-based methods Zahra Sadri

More information

Robot Localization based on Geo-referenced Images and G raphic Methods

Robot Localization based on Geo-referenced Images and G raphic Methods Robot Localization based on Geo-referenced Images and G raphic Methods Sid Ahmed Berrabah Mechanical Department, Royal Military School, Belgium, sidahmed.berrabah@rma.ac.be Janusz Bedkowski, Łukasz Lubasiński,

More information

Pedestrian Detection with Improved LBP and Hog Algorithm

Pedestrian Detection with Improved LBP and Hog Algorithm Open Access Library Journal 2018, Volume 5, e4573 ISSN Online: 2333-9721 ISSN Print: 2333-9705 Pedestrian Detection with Improved LBP and Hog Algorithm Wei Zhou, Suyun Luo Automotive Engineering College,

More information

Robot localization method based on visual features and their geometric relationship

Robot localization method based on visual features and their geometric relationship , pp.46-50 http://dx.doi.org/10.14257/astl.2015.85.11 Robot localization method based on visual features and their geometric relationship Sangyun Lee 1, Changkyung Eem 2, and Hyunki Hong 3 1 Department

More information

Postprint.

Postprint. http://www.diva-portal.org Postprint This is the accepted version of a paper presented at 14th International Conference of the Biometrics Special Interest Group, BIOSIG, Darmstadt, Germany, 9-11 September,

More information

Research on Robust Local Feature Extraction Method for Human Detection

Research on Robust Local Feature Extraction Method for Human Detection Waseda University Doctoral Dissertation Research on Robust Local Feature Extraction Method for Human Detection TANG, Shaopeng Graduate School of Information, Production and Systems Waseda University Feb.

More information

Mouse Pointer Tracking with Eyes

Mouse Pointer Tracking with Eyes Mouse Pointer Tracking with Eyes H. Mhamdi, N. Hamrouni, A. Temimi, and M. Bouhlel Abstract In this article, we expose our research work in Human-machine Interaction. The research consists in manipulating

More information

Ensemble of Bayesian Filters for Loop Closure Detection

Ensemble of Bayesian Filters for Loop Closure Detection Ensemble of Bayesian Filters for Loop Closure Detection Mohammad Omar Salameh, Azizi Abdullah, Shahnorbanun Sahran Pattern Recognition Research Group Center for Artificial Intelligence Faculty of Information

More information

Adaptive Skin Color Classifier for Face Outline Models

Adaptive Skin Color Classifier for Face Outline Models Adaptive Skin Color Classifier for Face Outline Models M. Wimmer, B. Radig, M. Beetz Informatik IX, Technische Universität München, Germany Boltzmannstr. 3, 87548 Garching, Germany [wimmerm, radig, beetz]@informatik.tu-muenchen.de

More information

Human Object Classification in Daubechies Complex Wavelet Domain

Human Object Classification in Daubechies Complex Wavelet Domain Human Object Classification in Daubechies Complex Wavelet Domain Manish Khare 1, Rajneesh Kumar Srivastava 1, Ashish Khare 1(&), Nguyen Thanh Binh 2, and Tran Anh Dien 2 1 Image Processing and Computer

More information

The Population Density of Early Warning System Based On Video Image

The Population Density of Early Warning System Based On Video Image International Journal of Research in Engineering and Science (IJRES) ISSN (Online): 2320-9364, ISSN (Print): 2320-9356 Volume 4 Issue 4 ǁ April. 2016 ǁ PP.32-37 The Population Density of Early Warning

More information

A Street Scene Surveillance System for Moving Object Detection, Tracking and Classification

A Street Scene Surveillance System for Moving Object Detection, Tracking and Classification A Street Scene Surveillance System for Moving Object Detection, Tracking and Classification Huei-Yung Lin * and Juang-Yu Wei Department of Electrical Engineering National Chung Cheng University Chia-Yi

More information

Traffic Sign Localization and Classification Methods: An Overview

Traffic Sign Localization and Classification Methods: An Overview Traffic Sign Localization and Classification Methods: An Overview Ivan Filković University of Zagreb Faculty of Electrical Engineering and Computing Department of Electronics, Microelectronics, Computer

More information

Integrated Vehicle and Lane Detection with Distance Estimation

Integrated Vehicle and Lane Detection with Distance Estimation Integrated Vehicle and Lane Detection with Distance Estimation Yu-Chun Chen, Te-Feng Su, Shang-Hong Lai Department of Computer Science, National Tsing Hua University,Taiwan 30013, R.O.C Abstract. In this

More information

Estimation of Camera Motion with Feature Flow Model for 3D Environment Modeling by Using Omni-Directional Camera

Estimation of Camera Motion with Feature Flow Model for 3D Environment Modeling by Using Omni-Directional Camera Estimation of Camera Motion with Feature Flow Model for 3D Environment Modeling by Using Omni-Directional Camera Ryosuke Kawanishi, Atsushi Yamashita and Toru Kaneko Abstract Map information is important

More information

Toward Retail Product Recognition on Grocery Shelves

Toward Retail Product Recognition on Grocery Shelves Toward Retail Product Recognition on Grocery Shelves Gül Varol gul.varol@boun.edu.tr Boğaziçi University, İstanbul, Turkey İdea Teknoloji Çözümleri, İstanbul, Turkey Rıdvan S. Kuzu ridvan.salih@boun.edu.tr

More information

AR Cultural Heritage Reconstruction Based on Feature Landmark Database Constructed by Using Omnidirectional Range Sensor

AR Cultural Heritage Reconstruction Based on Feature Landmark Database Constructed by Using Omnidirectional Range Sensor AR Cultural Heritage Reconstruction Based on Feature Landmark Database Constructed by Using Omnidirectional Range Sensor Takafumi Taketomi, Tomokazu Sato, and Naokazu Yokoya Graduate School of Information

More information

Visual Odometry. Features, Tracking, Essential Matrix, and RANSAC. Stephan Weiss Computer Vision Group NASA-JPL / CalTech

Visual Odometry. Features, Tracking, Essential Matrix, and RANSAC. Stephan Weiss Computer Vision Group NASA-JPL / CalTech Visual Odometry Features, Tracking, Essential Matrix, and RANSAC Stephan Weiss Computer Vision Group NASA-JPL / CalTech Stephan.Weiss@ieee.org (c) 2013. Government sponsorship acknowledged. Outline The

More information

HOG-based Pedestriant Detector Training

HOG-based Pedestriant Detector Training HOG-based Pedestriant Detector Training evs embedded Vision Systems Srl c/o Computer Science Park, Strada Le Grazie, 15 Verona- Italy http: // www. embeddedvisionsystems. it Abstract This paper describes

More information

URBAN STRUCTURE ESTIMATION USING PARALLEL AND ORTHOGONAL LINES

URBAN STRUCTURE ESTIMATION USING PARALLEL AND ORTHOGONAL LINES URBAN STRUCTURE ESTIMATION USING PARALLEL AND ORTHOGONAL LINES An Undergraduate Research Scholars Thesis by RUI LIU Submitted to Honors and Undergraduate Research Texas A&M University in partial fulfillment

More information

Improving Vision-Based Distance Measurements using Reference Objects

Improving Vision-Based Distance Measurements using Reference Objects Improving Vision-Based Distance Measurements using Reference Objects Matthias Jüngel, Heinrich Mellmann, and Michael Spranger Humboldt-Universität zu Berlin, Künstliche Intelligenz Unter den Linden 6,

More information

CS 223B Computer Vision Problem Set 3

CS 223B Computer Vision Problem Set 3 CS 223B Computer Vision Problem Set 3 Due: Feb. 22 nd, 2011 1 Probabilistic Recursion for Tracking In this problem you will derive a method for tracking a point of interest through a sequence of images.

More information

Skin and Face Detection

Skin and Face Detection Skin and Face Detection Linda Shapiro EE/CSE 576 1 What s Coming 1. Review of Bakic flesh detector 2. Fleck and Forsyth flesh detector 3. Details of Rowley face detector 4. Review of the basic AdaBoost

More information

REAL-TIME ROAD SIGNS RECOGNITION USING MOBILE GPU

REAL-TIME ROAD SIGNS RECOGNITION USING MOBILE GPU High-Performance Сomputing REAL-TIME ROAD SIGNS RECOGNITION USING MOBILE GPU P.Y. Yakimov Samara National Research University, Samara, Russia Abstract. This article shows an effective implementation of

More information

Vision-based Bicyclist Detection and Tracking for Intelligent Vehicles

Vision-based Bicyclist Detection and Tracking for Intelligent Vehicles Vision-based Bicyclist Detection and Tracking for Intelligent Vehicles Hyunggi Cho, Paul E. Rybski and Wende Zhang Abstract This paper presents a vision-based framework for intelligent vehicles to detect

More information

Vision Based Parking Space Classification

Vision Based Parking Space Classification 1 Vision Based Parking Space Classification Ananth Nallamuthu, Sandeep Lokala, Department of ECE, Clemson University. Abstract The problem of Vacant Parking space detection from static images using computer

More information

Scene Text Detection Using Machine Learning Classifiers

Scene Text Detection Using Machine Learning Classifiers 601 Scene Text Detection Using Machine Learning Classifiers Nafla C.N. 1, Sneha K. 2, Divya K.P. 3 1 (Department of CSE, RCET, Akkikkvu, Thrissur) 2 (Department of CSE, RCET, Akkikkvu, Thrissur) 3 (Department

More information

arxiv: v1 [cs.cv] 28 Sep 2018

arxiv: v1 [cs.cv] 28 Sep 2018 Camera Pose Estimation from Sequence of Calibrated Images arxiv:1809.11066v1 [cs.cv] 28 Sep 2018 Jacek Komorowski 1 and Przemyslaw Rokita 2 1 Maria Curie-Sklodowska University, Institute of Computer Science,

More information

Human Detection and Tracking for Video Surveillance: A Cognitive Science Approach

Human Detection and Tracking for Video Surveillance: A Cognitive Science Approach Human Detection and Tracking for Video Surveillance: A Cognitive Science Approach Vandit Gajjar gajjar.vandit.381@ldce.ac.in Ayesha Gurnani gurnani.ayesha.52@ldce.ac.in Yash Khandhediya khandhediya.yash.364@ldce.ac.in

More information

(a) (b) (c) Fig. 1. Omnidirectional camera: (a) principle; (b) physical construction; (c) captured. of a local vision system is more challenging than

(a) (b) (c) Fig. 1. Omnidirectional camera: (a) principle; (b) physical construction; (c) captured. of a local vision system is more challenging than An Omnidirectional Vision System that finds and tracks color edges and blobs Felix v. Hundelshausen, Sven Behnke, and Raul Rojas Freie Universität Berlin, Institut für Informatik Takustr. 9, 14195 Berlin,

More information

Human detection using histogram of oriented gradients. Srikumar Ramalingam School of Computing University of Utah

Human detection using histogram of oriented gradients. Srikumar Ramalingam School of Computing University of Utah Human detection using histogram of oriented gradients Srikumar Ramalingam School of Computing University of Utah Reference Navneet Dalal and Bill Triggs, Histograms of Oriented Gradients for Human Detection,

More information

Crowd Event Recognition Using HOG Tracker

Crowd Event Recognition Using HOG Tracker Crowd Event Recognition Using HOG Tracker Carolina Gárate Piotr Bilinski Francois Bremond Pulsar Pulsar Pulsar INRIA INRIA INRIA Sophia Antipolis, France Sophia Antipolis, France Sophia Antipolis, France

More information

EXAM SOLUTIONS. Image Processing and Computer Vision Course 2D1421 Monday, 13 th of March 2006,

EXAM SOLUTIONS. Image Processing and Computer Vision Course 2D1421 Monday, 13 th of March 2006, School of Computer Science and Communication, KTH Danica Kragic EXAM SOLUTIONS Image Processing and Computer Vision Course 2D1421 Monday, 13 th of March 2006, 14.00 19.00 Grade table 0-25 U 26-35 3 36-45

More information

CS 231A Computer Vision (Fall 2012) Problem Set 3

CS 231A Computer Vision (Fall 2012) Problem Set 3 CS 231A Computer Vision (Fall 2012) Problem Set 3 Due: Nov. 13 th, 2012 (2:15pm) 1 Probabilistic Recursion for Tracking (20 points) In this problem you will derive a method for tracking a point of interest

More information

CS4758: Moving Person Avoider

CS4758: Moving Person Avoider CS4758: Moving Person Avoider Yi Heng Lee, Sze Kiat Sim Abstract We attempt to have a quadrotor autonomously avoid people while moving through an indoor environment. Our algorithm for detecting people

More information

DPM Score Regressor for Detecting Occluded Humans from Depth Images

DPM Score Regressor for Detecting Occluded Humans from Depth Images DPM Score Regressor for Detecting Occluded Humans from Depth Images Tsuyoshi Usami, Hiroshi Fukui, Yuji Yamauchi, Takayoshi Yamashita and Hironobu Fujiyoshi Email: usami915@vision.cs.chubu.ac.jp Email:

More information

A Two-Stage Template Approach to Person Detection in Thermal Imagery

A Two-Stage Template Approach to Person Detection in Thermal Imagery A Two-Stage Template Approach to Person Detection in Thermal Imagery James W. Davis Mark A. Keck Ohio State University Columbus OH 432 USA {jwdavis,keck}@cse.ohio-state.edu Abstract We present a two-stage

More information

3D Environment Measurement Using Binocular Stereo and Motion Stereo by Mobile Robot with Omnidirectional Stereo Camera

3D Environment Measurement Using Binocular Stereo and Motion Stereo by Mobile Robot with Omnidirectional Stereo Camera 3D Environment Measurement Using Binocular Stereo and Motion Stereo by Mobile Robot with Omnidirectional Stereo Camera Shinichi GOTO Department of Mechanical Engineering Shizuoka University 3-5-1 Johoku,

More information

Pedestrian Detection Using Correlated Lidar and Image Data EECS442 Final Project Fall 2016

Pedestrian Detection Using Correlated Lidar and Image Data EECS442 Final Project Fall 2016 edestrian Detection Using Correlated Lidar and Image Data EECS442 Final roject Fall 2016 Samuel Rohrer University of Michigan rohrer@umich.edu Ian Lin University of Michigan tiannis@umich.edu Abstract

More information

Segmentation and Tracking of Partial Planar Templates

Segmentation and Tracking of Partial Planar Templates Segmentation and Tracking of Partial Planar Templates Abdelsalam Masoud William Hoff Colorado School of Mines Colorado School of Mines Golden, CO 800 Golden, CO 800 amasoud@mines.edu whoff@mines.edu Abstract

More information

Adaptive Gesture Recognition System Integrating Multiple Inputs

Adaptive Gesture Recognition System Integrating Multiple Inputs Adaptive Gesture Recognition System Integrating Multiple Inputs Master Thesis - Colloquium Tobias Staron University of Hamburg Faculty of Mathematics, Informatics and Natural Sciences Technical Aspects

More information

Sensor Fusion-Based Parking Assist System

Sensor Fusion-Based Parking Assist System Sensor Fusion-Based Parking Assist System 2014-01-0327 Jaeseob Choi, Eugene Chang, Daejoong Yoon, and Seongsook Ryu Hyundai & Kia Corp. Hogi Jung and Jaekyu Suhr Hanyang Univ. Published 04/01/2014 CITATION:

More information

High-Level Fusion of Depth and Intensity for Pedestrian Classification

High-Level Fusion of Depth and Intensity for Pedestrian Classification High-Level Fusion of Depth and Intensity for Pedestrian Classification Marcus Rohrbach 1,3, Markus Enzweiler 2 and Dariu M. Gavrila 1,4 1 Environment Perception, Group Research, Daimler AG, Ulm, Germany

More information

Vision-based Bicyclist Detection and Tracking for Intelligent Vehicles

Vision-based Bicyclist Detection and Tracking for Intelligent Vehicles Vision-based Bicyclist Detection and Tracking for Intelligent Vehicles Hyunggi Cho Paul E. Rybski Wende Zhang CMU-RI-TR-10-11 January 2010 Robotics Institute Carnegie Mellon University Pittsburgh, Pennsylvania

More information

Monocular Vision Based Autonomous Navigation for Arbitrarily Shaped Urban Roads

Monocular Vision Based Autonomous Navigation for Arbitrarily Shaped Urban Roads Proceedings of the International Conference on Machine Vision and Machine Learning Prague, Czech Republic, August 14-15, 2014 Paper No. 127 Monocular Vision Based Autonomous Navigation for Arbitrarily

More information

Tri-modal Human Body Segmentation

Tri-modal Human Body Segmentation Tri-modal Human Body Segmentation Master of Science Thesis Cristina Palmero Cantariño Advisor: Sergio Escalera Guerrero February 6, 2014 Outline 1 Introduction 2 Tri-modal dataset 3 Proposed baseline 4

More information

Building Reliable 2D Maps from 3D Features

Building Reliable 2D Maps from 3D Features Building Reliable 2D Maps from 3D Features Dipl. Technoinform. Jens Wettach, Prof. Dr. rer. nat. Karsten Berns TU Kaiserslautern; Robotics Research Lab 1, Geb. 48; Gottlieb-Daimler- Str.1; 67663 Kaiserslautern;

More information

An Introduction to Content Based Image Retrieval

An Introduction to Content Based Image Retrieval CHAPTER -1 An Introduction to Content Based Image Retrieval 1.1 Introduction With the advancement in internet and multimedia technologies, a huge amount of multimedia data in the form of audio, video and

More information

Jo-Car2 Autonomous Mode. Path Planning (Cost Matrix Algorithm)

Jo-Car2 Autonomous Mode. Path Planning (Cost Matrix Algorithm) Chapter 8.2 Jo-Car2 Autonomous Mode Path Planning (Cost Matrix Algorithm) Introduction: In order to achieve its mission and reach the GPS goal safely; without crashing into obstacles or leaving the lane,

More information

Accurate Motion Estimation and High-Precision 3D Reconstruction by Sensor Fusion

Accurate Motion Estimation and High-Precision 3D Reconstruction by Sensor Fusion 007 IEEE International Conference on Robotics and Automation Roma, Italy, 0-4 April 007 FrE5. Accurate Motion Estimation and High-Precision D Reconstruction by Sensor Fusion Yunsu Bok, Youngbae Hwang,

More information

IMAGE RETRIEVAL USING VLAD WITH MULTIPLE FEATURES

IMAGE RETRIEVAL USING VLAD WITH MULTIPLE FEATURES IMAGE RETRIEVAL USING VLAD WITH MULTIPLE FEATURES Pin-Syuan Huang, Jing-Yi Tsai, Yu-Fang Wang, and Chun-Yi Tsai Department of Computer Science and Information Engineering, National Taitung University,

More information

Detecting People in Images: An Edge Density Approach

Detecting People in Images: An Edge Density Approach University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 27 Detecting People in Images: An Edge Density Approach Son Lam Phung

More information

HUMAN POSTURE DETECTION WITH THE HELP OF LINEAR SVM AND HOG FEATURE ON GPU

HUMAN POSTURE DETECTION WITH THE HELP OF LINEAR SVM AND HOG FEATURE ON GPU International Journal of Computer Engineering and Applications, Volume IX, Issue VII, July 2015 HUMAN POSTURE DETECTION WITH THE HELP OF LINEAR SVM AND HOG FEATURE ON GPU Vaibhav P. Janbandhu 1, Sanjay

More information

SUMMARY: DISTINCTIVE IMAGE FEATURES FROM SCALE- INVARIANT KEYPOINTS

SUMMARY: DISTINCTIVE IMAGE FEATURES FROM SCALE- INVARIANT KEYPOINTS SUMMARY: DISTINCTIVE IMAGE FEATURES FROM SCALE- INVARIANT KEYPOINTS Cognitive Robotics Original: David G. Lowe, 004 Summary: Coen van Leeuwen, s1460919 Abstract: This article presents a method to extract

More information

Deep Tracking: Biologically Inspired Tracking with Deep Convolutional Networks

Deep Tracking: Biologically Inspired Tracking with Deep Convolutional Networks Deep Tracking: Biologically Inspired Tracking with Deep Convolutional Networks Si Chen The George Washington University sichen@gwmail.gwu.edu Meera Hahn Emory University mhahn7@emory.edu Mentor: Afshin

More information

Recognition of a Predefined Landmark Using Optical Flow Sensor/Camera

Recognition of a Predefined Landmark Using Optical Flow Sensor/Camera Recognition of a Predefined Landmark Using Optical Flow Sensor/Camera Galiev Ilfat, Alina Garaeva, Nikita Aslanyan The Department of Computer Science & Automation, TU Ilmenau 98693 Ilmenau ilfat.galiev@tu-ilmenau.de;

More information

A Keypoint Descriptor Inspired by Retinal Computation

A Keypoint Descriptor Inspired by Retinal Computation A Keypoint Descriptor Inspired by Retinal Computation Bongsoo Suh, Sungjoon Choi, Han Lee Stanford University {bssuh,sungjoonchoi,hanlee}@stanford.edu Abstract. The main goal of our project is to implement

More information

Monitoring surrounding areas of truck-trailer combinations

Monitoring surrounding areas of truck-trailer combinations Monitoring surrounding areas of truck-trailer combinations Tobias Ehlgen 1 and Tomas Pajdla 2 1 Daimler-Chrysler Research and Technology, Ulm tobias.ehlgen@daimlerchrysler.com 2 Center of Machine Perception,

More information