Abstract

Terrains are of great interest in flight simulators, geographic information systems and computer games. In computer graphics, terrain rendering is a special case because of their bulk. They cannot be handled as a single entity like other object models like teapots, cars and crates. Triangulated irregular networks of terrains are typically created by simplifying a dense representation. Such representations are popular in GIS and computational geometry. The recent trend in graphics is to use regular grid representations since they go well with today’s graphics hardware. We explore different representation techniques to render terrains in this thesis. We look into real-time rendering, editing, and physical interaction with external objects on terrains. We also present a representation for efficient rendering of spherical terrains. Apart from rendering terrains realistically, we develop a method to render terrains artistically with painterly abstraction as well. We create a system that exploits the power and flexibility of the modern GPUs to store, render, and manipulate terrains with minimal CPU involvement (CPU load < 4%). The central idea is to use a regular-grid representation, hierarchically divided in fixed size blocks/tiles that change in resolution. The potentially visible portion of the terrain is cached at the highest necessary resolution and is rendered from the GPU. The cache is updated with the viewerpoint. Lower resolutions used for farther areas of the terrain can be constructed from the cache on the GPU. Todays GPUs have a limited capability to generate geometry within itself. Thus, the CPU can send a light geometry template which is expanded to the triangles by GPU. The CPU performs a coarse culling of the tiles with the GPU performing fine culling. Our system enables the terrain to be modified procedurally or edited interactively on the GPU with no CPU involvement. The terrain can also interact with a large number of external objects in real-time with all the physics calculations done on the GPU. Terrains can also be mapped over a sphere for a planetary structure. However, terrains on sphere require a different representation due to the pole singularity of latitude-longitude representation. A 2D grid of height cannot be mapped directly on a sphere with uniform triangle count. Spheres can be rendered uniformly using Hierarchical Triangular Mesh (HTM) but the representation does not fit with 2D grid of heightmaps. We present a unified representation of HTM and clipmapping (flat terrain rendering technique) to render spherical terrains. Our representation works at any distance from the planet/sphere without any scripted work arounds. The regular nature of terrain data also enables us to render the samples in a required order with no overhead of sorting. This introduces the possibility of applications which require sorted ordered vi vii triangles. We explore non photo realisitic rendering of terrains. Artistic painterly appearance and the impression of terrains is created by effectively rendering several translucent brush strokes in a back to front order. The strokes are located in 3D space for frame-to-frame coherence during animation. The strokes are oriented along the slope of terrain analogous to the way artists paint on canvas. We use shaders to render strokes in real-time. A level of detail scheme is used to maintain a uniform stroke density in screen space. Various styles can be achieved with different stroke variations. We achieve real-time painterly rendering with a combination of object space positioning and image space rendering of strokes.

Abstract

The amount of multimedia content produced and made available on Internet and in professional and personal collections is constantly growing. Equally increasing are the needs in terms of efficient and effective ways to manage it. This has led to a great amount of research into content based retrieval and visual recognition. In this thesis, we focus on efficient visual content analysis in images and videos. Efficiency has emerged as one of the key issues with increase in quantity of data. Understanding of a visual content has several aspects associated with it. One can concentrate on recognizing the inherent characteristics of image (independent or from a video) like objects, scene and context. Searching for a sequence of images based on similarity or characterizing the video based on its visual content could be some other aspects. We investigate three different approaches for visual content analysis in this thesis. In the first, we target the detection and classification of different object and scene classes in images and videos. The task of classification is to predict the presence of an object or a specific scene of interest in the test image. Object detection further involves localizing each instance of the object present. We do extensive experimentation over very large and challenging datasets with large number of object and scene categories in it. Our detection as well as classification are based on Random Forest combined with combinations of different visual features describing shape, appearance and color. We exploited the computational efficiency in both training and testing, and other properties of Random Forest for detection and classification. We also proposed enhancements over our baseline model of object detector. Our main contribution here is that we achieve fast object detection with accuracy comparable to the state of art. The second approach is based on processing continuous stream of videos to detect video segments of interest. Our method is example-based where visual content to be detected or filtered is characterized by a set of examples available apriori. We approach the problem of video processing in a manner complimentary to that of video retrieval. We begin with a set of examples (used as queries in retrieval) and index them in the database. The larger video collection, which needs to be processed, is unseen during the off-line indexing phase. We propose an architecture based on trie data structure and bag of words model to simultaneously match multiple example videos in the database with the input large video stream. We demonstrate the application of our architecture for the task of content based copy detection (CBCD). xiii xiv In our third and final approach we apply pattern mining algorithms in videos to characterize the visual content. They are derived out of data mining schemes for efficient analysis of the content in video databases. Two different video mining schemes are employed; both aimed at detecting frequent and representative patterns. For one of our mining approaches, we use an efficient frequent pattern mining algorithm over a quantized feature space. Our second approach uses random forest to represent video data as sequences, and mine the frequent sequences. We experiment on broadcast news videos to detect what we define as video stop-words and extract the contents which are more important such as breaking news. We are also able to characterize the movie videos by automatically identifying the characteristic scenes and main actors of the movie. The ideas proposed in the thesis have been implemented and validated with extensive experimental results. We demonstrate the accuracy, efficiency and scalability of all the three approaches over large and standrad datasets like VOC PASCAL, TRECVID, MUSCLE-VCD as well as movie and news datasets.

Abstract

Information Retrieval systems like Web search engines are often used by people to find information of their interest. Despite the success of these systems, they are not sophisticated enough to provide precise information for users requests. To address this problem, complex Information Access systems called Question Answering systems have been developed. They aim at finding exact answers to natural language questions in a large collection of documents (such as World Wide Web). One of the most obvious limitations in the performance of many Question Answering systems is their inability to find text passages where candidate answers can be found. Earlier research on the poor performance of passage retrieval highlighted the terminological gap problem i.e., passages holding the answer to a question have semantic alterations of original terms in the question. In this thesis, we proposed two different techniques to reduce this problem. Query expansion is a widely used technique in Information Retrieval to reduce the terminological gap problem. First, we present a novel passage retrieval methodology which expands the query inherently. This methodology leverages Statistical Machine Translation model for Information Retrieval to retrieve a ranked set of passages given a question. The retrieval within this model includes two steps: estimation and ranking. In the estimation step, multiple translation models are constructed using a statistical alignment model. We perceive each such translation model as an answer type profile. During ranking, based on the answer type of the question its corresponding answer type profile is used to retrieve relevant passages. Our experimental analysis on the performance of this retrieval methodology showed significant improvements over different standard retrieval models including TFIDF, Okapi BM25, Indri and KL-divergence. We found that simple statistical alignment models like IBM model 1 are more suited for passage retrieval. We also showed that our methodology addresses the problems of synonymy and polysemy. Previous studies on explicit query expansion methods like pseudo relevance feedback, and methods based on external knowledge sources likeWordNet,Wikipedia or Web have shown to improve the performance of Information Retrieval systems. We proposed a novel query expansion method using Wikipedia. Our methodology uses text content, category structure, and link structure of Wikipedia to generate a set of terms semantically related to the question. As Boolean model allows a finegrained control over query expansion, these semantically related terms are added to the original query to form an expanded Boolean query. Our experimental analysis on the performance of these expanded queries on Lucene, an open source retrieval system, showed significant improvements over seed queries. We also analyzed the performance of expanded queries based on different scoring methods utilized in selecting query expansion terms and for different query expansion lengths. Adding to the above contributions which focused on reducing the terminological gap between query and passage, we explored the necessity of passage priors in ranking passages. Document Retrieval assumes that a document is independent of its relevance, and non-relevance. The same assumption is being carried forward in passage retrieval systems in the context of Question Answering. We relax this assumption and explore the necessity of passage priors being relevant and non relevant in ranking passages given a query. We describe a mutual information measure for estimating these priors and a simple method for identifying relevant and non-relevant text to a question using the Web and AQUAINT corpus as information sources. Our experimental analysis of using passage priors as a re-ranking step on top of language models including Indri and KL-divergence showed that passage priors are necessary in ranking passages.

Abstract

Online shopping has become a very popular web application in the recent times and received lot of attention from both consumers and retail merchants. With this growing popularity, lots of information is generated on many products to assist users across the globe. This has resulted in information explosion with lots of information for the consumers to digest. This has created need for techniques to identify useful and relevant content for the consumers from the huge amounts available. In this thesis, we have explored application of information extraction techniques for extracting relevant product information. We have focused on the following subproblems in the context of products: Attribute Extraction, Attribute Ranking. We have explored algorithms to automatically extract product attributes from text descriptions. We have come up with unsupervised methods which do not require any domain specific information for extraction. Our algorithms are based on the hypothesis that attributes should repeat across descriptions. We have explored two clustering methods in this work for extraction : Noun Phrase(NP) Clustering and Word Clustering. In the first method, clusters of noun phrases are computed so that all the phrases describing an attribute are grouped together and a representative attribute is extracted from each cluster. In the second method, we cluster words appearing in the descriptions such that all the words related to an attribute are grouped together in one cluster. We construct a graph from the word occurrences in descriptions and compute word clusters using a graph clustering algorithm. Attributes are extracted from word clusters using word associations inside a cluster. We have throughly evaluated our methods by conducting various experiments. Our experiments show that the methods are robust and extract product attributes accurately. We have also compared the two algorithms and presented an analysis on the trade offs in the two approaches. We have defined attribute ranking in the context of product comparison and came up new features for ranking the attributes. We have also designed a new kind of summaries called Comparative Summaries that delivers majority of the comparable content found in an input set of documents. We have presented how comparative summaries could be generated for products using our attribute extraction and ranking algorithms. We have carried out experiments to evaluate the performance of our ranking algorithm and showed that it is effective in identifying useful attributes.

Abstract

The MapReduce paradigm has become a popular way of expressing distributed data processing problems that need to deal with large amount of data. MapReduce is used by a number of organizations worldwide for diverse tasks such as application log processing, user behavior analysis, processing scientific data, web crawling and indexing etc. Many organizations utilize cloud computing services to acquire resources needed to process MapReduce jobs on demand. As a result of the popularity of using MapReduce in the cloud, cloud services such as the Amazon Elastic MapReduce have also become available. The Apache Hadoop framework is the leading open source implementation of the MapReduce model. It also provides a distributed file system (HDFS or the Hadoop Distributed File System). Hadoop is also the most popular MapReduce framework, and is being used by more than 75 organizations worldwide. It is designed to be scalable and robust, and runs on small single node clusters, to very large clusters containing several thousand compute and storage nodes. Despite the popularity and stability of Hadoop, it presents many opportunities for researching new resource management algorithms. Admission Control, Task Assignment and Scheduling, Data Local Execution, Speculative Execution and Replica placement are some of the key challenges involved in resource management in Hadoop. Resource management in Hadoop is complicated by the fact that the resources available are dynamic in nature because of frequent node failures. In this thesis, we approach two of the above problems: Admission Control and Task Assignment. Admission Control is the problem of deciding if a new job submission should be accepted for execution on a MapReduce cluster. An admission controller, or the module that handles admission control should make sure that jobs accepted into the cluster do not overload resources. Another important requirement is that the accepted jobs should try to maximize the utility of the service provider, which in this case is the Hadoop cluster owner. We propose an admission control algorithm that selects incoming jobs based on the principle of expected utility hypothesis. Based on this principle, the admission controller always chooses jobs that are expected to provide maximum utility after their successful completion. To predict whether or not the job will be successful, we observe the effects of past decisions made under similar situations and use this knowledge to predict outcome of incoming jobs. For this purpose we use a Nave Bayes classifier that labels incoming jobs as potentially successful or potentially unsuccessful. Out of the jobs that are labeled potentially successful, we chose the job that has the maximum value of utility function at the time of job submission. If none of the incoming jobs are labeled as potentially successful, we do not admit any job into the cluster. Utility functions are supplied by users and express the value earned by users after successful completion of jobs as a function of time taken to complete their requests. We consider a job submission to be successful if it does not overload resources on the cluster. Once the effects of the admission are observed, we update the classifier so that this experience is used while performing next decision. The next problem that we approach is that of Task Assignment. Each MapReduce job is subdivided into a number of map and reduce tasks. The tasks of a job are executed concurrently. Task assignment is the problem of deciding which task should be allocated to a node in the MapReduce cluster. The task assignment algorithm should consider the current state of resources at the concerned node as well as the resources requirements of tasks in the schedulers queue. Given this information, the task assignment algorithm should choose a task that maximizes resource utilization on the concerned node while still making sure that the node is not overloaded. We used a similar approach as that used in the admission controller for task assignment as well. While choosing a task, we first classify the queued tasks into two categories, good and bad. Good tasks do not overload resources at a node during their execution. From the tasks that are labeled good, we choose the task that has the maximum utility, which in this case is the priority of a task. Again, if none of the tasks are classified as good, we do not assign any task to the concerned node. Task priority is set by the cluster administrator, and can be used for policy enforcement. We thus decouple task assignment from policy enforcement. Once a task is assigned, we observe its effects, and if it results in overloading of resources at the concerned node, we conclude that the task assignment decision was incorrect and update the classifier accordingly. This makes sure that chances of assignments that result into overloading are reduced over time. Overload rules are specified by system administrator.

Abstract

Analog to Digital Converter (ADC) is a fundamental component in digital signal processing systems. For high-speed applications, a flash ADC is often used. Resolution, speed, and power consumption are the three key parameters for an analog-to-digital converter (ADC), which cannot be changed once an ADC is designed. While one can use only 8-bit precision from a 10-bit ADC. Such an application is non-optimal, resulting in slower speed and extra power consumption due to full 10-bit internal operation. Flash ADCs architectures commonly use high speed conversion that requires many analog comparators, which in turn cause higher complexity and power dissipation. Moreover, the accuracy of dividing resistors requires high value for reference voltage if the converting resolution is high. In this thesis, a novel architecture for flash ADC is proposed. In this architecture comparators of conventional flash ADC are replaced with CMOS inverters whose threshold can be varied dynamically. A novel peak-detector circuit is employed to achieve variable resolution for the ADC as well as to switch the unused parallel inverters to standby mode. The ADC is capable of operating at 4-bit, 6-bit, and 8-bit precision at a supply voltage of 2.5V, it consumes 21.5mW at 8-bit, 9mW at 6-bit and 3.5mW at 4-bit resolution. The sampling frequency ranges from 0.8 to 1.6 GSPS, and the ADC has a DNL<±0.4LSB, INL<±0.36LSB, SNR of 47dB and SNDR of 46.3dB for 8-bit operation. The proposed ‘inverter-based’ flash ADC operating at 8-bit precision and conventional 8-bit comparator-based flash ADC have been designed, compared and verified for post layout simulations in standard 65nm CMOS technology using thick gate process. It is observed that linear reduction in resolution leads to exponential reduction in power.

Abstract

Pattern discovery is one of the fundamental problems in bioinformatics and algorithms from computer science have been widely used for identifying biological patterns. The assumption behind pattern discovery approaches is that a pattern that occurs often enough in biological sequences/structures or conserved across organisms is expected to play a role in defining the respective sequence’s or structure’s functional behavior and/or evolutionary relationships. Typical examples of pattern recognition at the sequence level involves identifying repeats, conserved regulatory sequences, protein-coding regions, horizontally transferred regions, motifs, profiles, etc. Similarly at the structure level pattern recognition involves identifying domains, structural motifs and repeats, etc. In this thesis we have addressed to problems Algorithms from computer science have been widely used for identifying biological patterns. In this thesis we have addressed two problems of pattern identification: the first one is at the structure level and involves identifying structural tandem repeats using graph theoretic approaches. The other pattern recognition problem addressed here is at the genomic level and involves identifying horizontally transferred regions. A web-based tool, IGIPT, incorporating a number of measures based on anomalies in nucleotide compositions has been developed.
1. Graph Theoretic Analysis of Protein Structures
Recently much attention is being focused on the analysis of the three dimensional structure of proteins using graph theory. Here, the protein structure is defined as a graph with nodes defined as the amino acids on the polypeptide backbone chain and edges between amino acids which attempt to capture not only covalent but also non-covalent interactions. A wide variety of problems dealing with protein structure such as protein fold recognition, active site identification, detection of functionally and structurally important amino acids, domain identification, etc. have been addressed using graph theory. Here, in this study we propose to use graph properties and their spectral analysis for identifying tandemly repeated patterns in protein structures. Specifically, we focus on identifying structural repeats in proteins which are not easily detectable at the sequence level because of low conservation and high divergence between independent repeated units. Thus, analysis at the structure level would be more reliable. Also, structural analysis of the repeat regions may give some insight to the possible function of the repeat assembly. This has been carried out by constructing protein contact graphs from its 3-dimensional structure and using tools from graph theory to investigate the topology of the 3-dimensional structure of the protein. The proposed approach when implemented on a number of proteins belonging to different repeat families in Pfam database, viz., Armidallo, Spectrin, HEAT, Ankyrin, and Leucine-rich repeats and on a number of proteins from structural repeats in PROPEAT database show promising results. Compared to the self structure-comparison approach of PROPEAT database, the graph theory provides a very elegant and efficient method of identifying repeats in protein structures. Since this approach can be easily automated, one can easily carry out a comparison of multiple copies of repeats between homologues. This would help in gaining insight into the mechanisms of evolution of these proteins by observing the conservation of the copy number or gain/loss of repeating units.
2. Integrated Genomic Island Prediction Tool
Numerous biological events are responsible for the evolution to occur such as gene conversions, rearrangements (e.g., inversion or translocation), deletions, and insertions of foreign DNA (e.g., plasmid integration, transposition). Horizontal Gene Transfer (HGT) is one of the major events responsible for causing significant alterations in the genome composition. This event is more strongly supported in case of prokaryotes and archaea as compared to eukaryotes. Horizontally transferred regions generally display abnormal nucleotide compositions and measures that capture these anomalies are used for their identification. We have developed a web-based tool for identifying genomic islands in prokaryotic genomes (IGIPT) in which various measures aimed at capturing anomalous nucleotide compositions have been incorporated, viz., GC content (at the genome level and at individual codon positions), dinucleotide biases, biases in codon and amino acid usage and k-mer distribution (k=2 to 6). The objective of developing such a tool was to provide all nucleotide based measures on a single platform since it has been observed that no single measure is sufficient in identifying GIs. Also, when more than one measure predicts the same region in the genome, it increases the confidence in the prediction. IGIPT has been tested on a set of prokaryotic geno

Abstract

Human visual system is a pattern seeker of enormous power and subtlety. We not only can see things clearly, but are also capable of describing them with precision and remembering them for a long time. Having these capabilities had a major impact on our sustenance, survival and perpetuation as species. Although computers can perform a variety of tasks that are beyond human capability because of speed, complexity, or dangerous environments, attempts to replicate human perceptual abilities have been strikingly inferior, even for the visual tasks that people consider extremely simple. In this thesis, we advance the research in the field of human computation by leveraging human perceptual abilities to solve problems that computers alone cannot effectively solve. In particular, we address two important problems: user authentication, and image annotation. User authentication has issues in both security and usability. For example, passwords are either ‘secure but difficult to remember’ or ‘memorable but not secure’, when by definition, they needs be both secure and memorable. Graphical passwords are viable alternative to text passwords since they are based on proven human ability to recognize and remember images, coupled with the larger password space offered by images. In this thesis, we propose and evaluate, Marasim, a novel Jigsaw based graphical authentication scheme, using Tagging. Marasim is aimed at achieving the security of system chosen images with the memorability of self chosen images. Empirical studies of Marasim provide evidence of increased memorability, usability and security. Additionally, we examine the manual image annotation problem. Recently there have been a number of attempts to lure humans into annotation process. Notable examples are interactive games like ESP, and social tagging like Flickr. However, we found that extant methods in their present form are inadequate to result in annotations of high quality. We therefore, introduce two intelligent system designs for semantic annotation of images in the form of a game and a CAPTCHA. First one is GoFish, a web variant of standard Go-Fish, a popular playing card game. While the other one is image recognition CAPTCHA, named iCAPTCHA. Behind both these designs is a strong emergent semantic theory that ensures superior annotations. Keywords: Human Computation, Perceptive Intelligence, Human in the loop system, Usable security, User Authentication, Graphical passwords, Tagging, Games with a Purpose, CAPTCHA.

Abstract

Need for better research robot platform and software development framework: The type of robots, sensor payload, etc varies with research problems and applications. A robot system with open and modular architecture enables addition of sensors, modification of overall system much easier, aid in speeding up the research work and customizing the robotic system as per requirements of the applications. Most of the commercial research robot platforms like Amigobot, Pioneer, Khepera offer limited options to modify the robot system. While this is good, in a way, in that it leads to research work in a common stream, there is a tendency for the research area to grow around the capabilities of robots. Cost of the robotic system, upgrading etc, also plays crucial role when there is limited funding. An affordable and open architecture robotic system will enable more researchers to work on actual robots instead of simulators. The complexity of the robot algorithms can be reduced by using sensor fusion of multiple sensors, to extract more accurate data. Computationally different algorithms can be developed by streamlining various modules to work in parallel. Unlike computers, robots work in highly dynamic environment; actuators and sensors are not accurate. This makes software development for robot far more complicated than developing software for computers. The features, capabilities and performance of different robot platforms are different, making it difficult to port software between different types of robots. Majority of the robotics software development is happening in universities and research labs, whose primary goal is to make scientific contributions in various areas of robotics. There is a need to integrate the existing work, in order to progress to higher level research problems. In order to handle large volume of data from multiple sensors, streamlined data management solutions are required. Features and advantages of our research robot platform: With the above goals in mind we have developed a new indoor research robotic system, which we describe in this thesis. It comprises of a differential drive wheel base, and an array of multiple ranging sensors which enable better data accuracy through sensor fusion. A digital compass provides additional robot heading information in addition to pose(x, y, θ), of the robot, from wheel encoder. The embedded controller is modularized into two distinct parts, a motion controller and a sensor controller. The motion controller is designed such that, the robot base comprising motors, encoders and wheels can be changed with very little modification in the embedded controller software. The robot is controlled by computer via ZigBee wireless communication link. Overall data exchange between computer and robot is structured such that frequency and volume of communication is reduced, due to less communication over head. This will enable more number of robots to simultaneously communicate in limited bandwidth. The design and architecture of this robotic system has been done keeping the requirements of robotics researchers and single and multi robot research platforms in mind. This robot at its current stage has most of the features required to be commercialized. It costs a fraction of similar commercial research robots, and provides better sensors and communication features. Features and advantages of robot software development framework This framework enables the algorithm to be divided into independent modules which can be run as separate threads or processes. Database tables are used to implement inter process communication (IPC). The output result of a module is written into a table and this data is used as input to some other modules. As database can be accessed over the network, processes can run on single or multiple computers. The processes are programming language and OS independent. Structured Query Language (SQL) provides various ways of accessing the data, which can enable us to design algorithms with computationally different approaches and reduce the burden of data management. As this framework supports robot software to run as separate processes, it is possible to run different robot control programs as separate processes and integrate those using tables as IPC.

Abstract

Scene interpretation is a fundamental task in both computer vision and robotic systems. We deal with two important aspects of scene interpretation, they are scene reconstruction and scene recognition. Scene reconstruction is determining 3D positions of world points and retrieving camera poses from images. It has several applications such as virtual building editing in computer aided architecture, video augmentation in film industry and planning and navigation in mobile robotics. Among several approaches to modeling the scene, we deal with piecewise planar modeling due to several advantages: Man-made environments are often piece-wise-planar, planar modeling has compact representation and this can be easily modified. We propose a convex optimization based, approach for piecewise planar reconstruction. We show that the task of reconstructing a piece-wise planar environment can be set in an L∞ based Homographic framework that iteratively computes scene plane and camera pose parameters. Instead of image points, the algorithm optimizes over inter-image homographies. The resultant objective function is minimized using Second Order Cone Programming algorithms. Apart from showing the convergence of the algorithm, we also empirically verify its robustness to error in initialization through various experiments on synthetic and real data. We intend this algorithm to be in between initialization approaches like decomposition methods and iterative non-linear minimization methods like Bundle Adjustment. Scene recognition in robotics, specifically terrain scene recognition is one of the fundamental tasks of autonomous navigation. Navigable terrains are examples of planar scenes. The goal of terrain recognition is to recognize various terrains that occur in urban and rural environments in an automated fashion. It has applications in various domains such as advanced driver assistance systems, remote sensing, etc. Various sensing modalities such as ladars, lasers, accelerometers, stereo cameras, omni-directional cameras or combination of them are used in literature. This thesis attacks the problem of scene interpretation using a single camera. This investigation is especially crucial since cameras are relatively low in cost, consume low power, light weight and have the potential to provide very rich information about the environment. Recent advances in computer vision, machine learning and improvements in hardware capabilities have greatly increased the scope of monocular camera, even in unstructured and real world environments. In this thesis, we start with empirical study of promising color, texture and their combination with classifiers such as Support Vector Machines (SVM) and Random Forests. We present comparison across features and classifiers. Then we present a monocular camera based terrain recognition scheme called Partition based classifier. The uniqueness of the proposed scheme is that it inherently incorporates spatial smoothness while segmenting an image, without the requirement of any additional post-processing. The algorithm is fast because it is build on top of a Random Forest classifier. The efficacy of the proposed solution can be seen as we reach low error rates on both our dataset and other publicly available datasets. Further partition classifier is extended to be online and adaptive. The new scheme consists of two underlying classifiers. One of which is learnt over bootstrapped or offline dataset, the second is another classifier that adapts to changes on the fly. Posterior probabilities of both the static and online classifiers are fused to assign the eventual label for the online image data. The online classifier learns at frequent intervals of time through a sparse and stable set of tracked patches, which makes it lightweight and real-time friendly. The learning which is acuted at frequent intervals during the sojourn significantly improves the performance of the classifier vis-a-vis a scheme that only uses the classifier learnt offline. The method finds immediate applications for outdoor autonomous driving where the classifier needs to be updated frequently based on what shows up recently on the terrain and without largely deviating from those learnt offline.