A. James Clark School of Engineering
Permanent URI for this communityhttp://hdl.handle.net/1903/1654
The collections in this community comprise faculty research works, as well as graduate theses and dissertations.
Browse
5 results
Search Results
Item DEEP NEURAL NETWORKS AND REGRESSION MODELS FOR OBJECT DETECTION AND POSE ESTIMATION(2016) Hara, Kota; Chellappa, Rama; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Estimating the pose, orientation and the location of objects has been a central problem addressed by the computer vision community for decades. In this dissertation, we propose new approaches for these important problems using deep neural networks as well as tree-based regression models. For the first topic, we look at the human body pose estimation problem and propose a novel regression-based approach. The goal of human body pose estimation is to predict the locations of body joints, given an image of a person. Due to significant variations introduced by pose, clothing and body styles, it is extremely difficult to address this task by a standard application of the regression method. Thus, we address this task by dividing the whole body pose estimation problem into a set of local pose estimation problems by introducing a dependency graph which describes the dependency among different body joints. For each local pose estimation problem, we train a boosted regression tree model and estimate the pose by progressively applying the regression along the paths in a dependency graph starting from the root node. Our next work is on improving the traditional regression tree method and demonstrate its effectiveness for pose/orientation estimation tasks. The main issues of the traditional regression training are, 1) the node splitting is limited to binary splitting, 2) the form of the splitting function is limited to thresholding on a single dimension of the input vector and 3) the best splitting function is found by exhaustive search. We propose a novel node splitting algorithm for regression tree training which does not have the issues mentioned above. The algorithm proceeds by first applying k-means clustering in the output space, conducting multi-class classification by support vector machine (SVM) and determining the constant estimate at each leaf node. We apply the regression forest that includes our regression tree models to head pose estimation, car orientation estimation and pedestrian orientation estimation tasks and demonstrate its superiority over various standard regression methods. Next, we turn our attention to the role of pose information for the object detection task. In particular, we focus on the detection of fashion items a person is wearing or carrying. It is clear that the locations of these items are strongly correlated with the pose of the person. To address this task, we first generate a set of candidate bounding boxes by using an object proposal algorithm. For each candidate bounding box, image features are extracted by a deep convolutional neural network pre-trained on a large image dataset and the detection scores are generated by SVMs. We introduce a pose-dependent prior on the geometry of the bounding boxes and combine it with the SVM scores. We demonstrate that the proposed algorithm achieves significant improvement in the detection performance. Lastly, we address the object detection task by exploring a way to incorporate an attention mechanism into the detection algorithm. Humans have the capability of allocating multiple fixation points, each of which attends to different locations and scales of the scene. However, such a mechanism is missing in the current state-of-the-art object detection methods. Inspired by the human vision system, we propose a novel deep network architecture that imitates this attention mechanism. For detecting objects in an image, the network adaptively places a sequence of glimpses at different locations in the image. Evidences of the presence of an object and its location are extracted from these glimpses, which are then fused for estimating the object class and bounding box coordinates. Due to the lack of ground truth annotations for the visual attention mechanism, we train our network using a reinforcement learning algorithm. Experiment results on standard object detection benchmarks show that the proposed network consistently outperforms the baseline networks that do not employ the attention mechanism.Item IP Geolocation in Metropolitan Areas(2011) Singh, Satinder Pal; Shayman, Mark; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)In this thesis, we propose a robust methodology to geolocate a target IP Address in a metropolitan area. We model the problem as a Pattern Recognition problem and present algorithms that can extract patterns and match them for inferring the geographic location of target's IP Address. The first algorithm is a relatively non-invasive method called Pattern Based Geolocation (PBG) which models the distribution of Round Trip Times (RTTs) to a target and matches them to that of the nearby landmarks to deduce the target's location. PBG builds Probability Mass Functions (PMFs) to model the distribution of RTTs. For comparing PMFs, we propose a novel `Shifted Symmetrized Divergence' distance metric which is a modified form of Kullback-Leibler divergence. It is symmetric as well as invariant to shifts. PBG algorithm works in almost stealth mode and leaves almost undetectable signature in network traffic. The second algorithm, Perturbation Augmented PBG (PAPBG), gives a higher resolution in the location estimate using additional perturbation traffic. The goal of this algorithm is to induce a stronger signature of background traffic in the vicinity of the target, and then detect it in the RTT sequences collected. At the cost of being intrusive, this algorithm improves the resolution of PBG by approximately 20-40%. We evaluate the performance of PBG and PAPBG on real data collected from 20 machines distributed over 700 square miles large Washington-Baltimore metropolitan area. We compare the performance of the proposed algorithms with existing measurement based geolocation techniques. Our experiments show that PBG shows marked improvements over current techniques and can geolocate a target IP address to within 2-4 miles of its actual location. And by sending an additional traffic in the network PAPBG improves the resolution to within 1-3 miles.Item Video Processing with Additional Information(2010) Ramachandran, Mahesh; Chellappa, Rama; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Cameras are frequently deployed along with many additional sensors in aerial and ground-based platforms. Many video datasets have metadata containing measurements from inertial sensors, GPS units, etc. Hence the development of better video processing algorithms using additional information attains special significance. We first describe an intensity-based algorithm for stabilizing low resolution and low quality aerial videos. The primary contribution is the idea of minimizing the discrepancy in the intensity of selected pixels between two images. This is an application of inverse compositional alignment for registering images of low resolution and low quality, for which minimizing the intensity difference over salient pixels with high gradients results in faster and better convergence than when using all the pixels. Secondly, we describe a feature-based method for stabilization of aerial videos and segmentation of small moving objects. We use the coherency of background motion to jointly track features through the sequence. This enables accurate tracking of large numbers of features in the presence of repetitive texture, lack of well conditioned feature windows etc. We incorporate the segmentation problem within the joint feature tracking framework and propose the first combined joint-tracking and segmentation algorithm. The proposed approach enables highly accurate tracking, and segmentation of feature tracks that is used in a MAP-MRF framework for obtaining dense pixelwise labeling of the scene. We demonstrate competitive moving object detection in challenging video sequences of the VIVID dataset containing moving vehicles and humans that are small enough to cause background subtraction approaches to fail. Structure from Motion (SfM) has matured to a stage, where the emphasis is on developing fast, scalable and robust algorithms for large reconstruction problems. The availability of additional sensors such as inertial units and GPS along with video cameras motivate the development of SfM algorithms that leverage these additional measurements. In the third part, we study the benefits of the availability of a specific form of additional information - the vertical direction (gravity) and the height of the camera both of which can be conveniently measured using inertial sensors, and a monocular video sequence for 3D urban modeling. We show that in the presence of this information, the SfM equations can be rewritten in a bilinear form. This allows us to derive a fast, robust, and scalable SfM algorithm for large scale applications. The proposed SfM algorithm is experimentally demonstrated to have favorable properties compared to the sparse bundle adjustment algorithm. We provide experimental evidence indicating that the proposed algorithm converges in many cases to solutions with lower error than state-of-art implementations of bundle adjustment. We also demonstrate that for the case of large reconstruction problems, the proposed algorithm takes lesser time to reach its solution compared to bundle adjustment. We also present SfM results using our algorithm on the Google StreetView research dataset, and several other datasets.Item Statistical/Geometric Techniques for Object Representation and Recognition(2009) Biswas, Soma; Chellappa, Rama; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Object modeling and recognition are key areas of research in computer vision and graphics with wide range of applications. Though research in these areas is not new, traditionally most of it has focused on analyzing problems under controlled environments. The challenges posed by real life applications demand for more general and robust solutions. The wide variety of objects with large intra-class variability makes the task very challenging. The difficulty in modeling and matching objects also vary depending on the input modality. In addition, the easy availability of sensors and storage have resulted in tremendous increase in the amount of data that needs to be processed which requires efficient algorithms suitable for large-size databases. In this dissertation, we address some of the challenges involved in modeling and matching of objects in realistic scenarios. Object matching in images require accounting for large variability in the appearance due to changes in illumination and view point. Any real world object is characterized by its underlying shape and albedo, which unlike the image intensity are insensitive to changes in illumination conditions. We propose a stochastic filtering framework for estimating object albedo from a single intensity image by formulating the albedo estimation as an image estimation problem. We also show how this albedo estimate can be used for illumination insensitive object matching and for more accurate shape recovery from a single image using standard shape from shading formulation. We start with the simpler problem where the pose of the object is known and only the illumination varies. We then extend the proposed approach to handle unknown pose in addition to illumination variations. We also use the estimated albedo maps for another important application, which is recognizing faces across age progression. Many approaches which address the problem of modeling and recognizing objects from images assume that the underlying objects are of diffused texture. But most real world objects exhibit a combination of diffused and specular properties. We propose an approach for separating the diffused and specular reflectance from a given color image so that the algorithms proposed for objects of diffused texture become applicable to a much wider range of real world objects. Representing and matching the 2D and 3D geometry of objects is also an integral part of object matching with applications in gesture recognition, activity classification, trademark and logo recognition, etc. The challenge in matching 2D/3D shapes lies in accounting for the different rigid and non-rigid deformations, large intra-class variability, noise and outliers. In addition, since shapes are usually represented as a collection of landmark points, the shape matching algorithm also has to deal with the challenges of missing or unknown correspondence across these data points. We propose an efficient shape indexing approach where the different feature vectors representing the shape are mapped to a hash table. For a query shape, we show how the similar shapes in the database can be efficiently retrieved without the need for establishing correspondence making the algorithm extremely fast and scalable. We also propose an approach for matching and registration of 3D point cloud data across unknown or missing correspondence using an implicit surface representation. Finally, we discuss possible future directions of this research.Item Adaptive Analysis and Processing of Structured Multilingual Documents(2006-01-19) Ma, Huanfeng; Chellappa, Rama; Doermann, David S; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Digital document processing is becoming popular for application to office and library automation, bank and postal services, publishing houses and communication management. In recent years, the demand for tools capable of searching written and spoken sources of multilingual information has increased tremendously, where the bilingual dictionary is one of the important resource to provide the required information. Processing and analysis of bilingual dictionaries brought up the challenges of dealing with many different scripts, some of which are unknown to the designer. A framework is presented to adaptively analyze and process structured multilingual documents, where adaptability is applied to every step. The proposed framework involves: (1) General word-level script identification using Gabor filter. (2) Font classification using the grating cell operator. (3) General word-level style identification using Gaussian mixture model. (4) An adaptable Hindi OCR based on generalized Hausdorff image comparison. (5) Retargetable OCR with automatic training sample creation and its applications to different scripts. (6) Bootstrapping entry segmentation, which segments each page into functional entries for parsing. Experimental results working on different scripts, such as Chinese, Korean, Arabic, Devanagari, and Khmer, demonstrate that the proposed framework can save human efforts significantly by making each phase adaptive.