"A meek endeavor to the triumph" by Sampath Jayarathna

Tuesday, December 14, 2010

Pictures of the Day - for some end term fun.......

Sri Lankan police!


Results of Love Marriage..........Its always the kids that suffer. His name is Zonky!!!!


Thursday, December 09, 2010

Reading #30: Tahuti: A Geometrical Sketch Recognition System for UML Class Diagrams

COMMENTS ON OTHERS:

            Jonathan 

SUMMARY

              Tahuti, a multi-stroke sketch recognition environment for class diagrams in UML where users can sketch the diagrams on a tablet or white- board in the same way they would on paper and the sketches are interpreted by the computer. Proposed system differs from graffiti-based approaches to this task in that it allows users to drawn an object as they would with pen and paper. The system recognizes objects based on their geometrical properties by examining the line segments’ angles, slopes, and other properties, rather than requiring the user to draw the objects in a pre-defined manner. Recognizing the objects by their geometrical properties gives users the freedom to sketch and edit diagrams as they would naturally, while maintaining a high level of recognition accuracy. Proposed system uses a multi-layer framework for sketch recognition. The multi-layer framework allows the system to recognize multi-stroke objects by their geometrical properties. The stages of the multi-layer recognition framework are: 1) Preprocessing 2) Selection 3) Recognition 4) Identification. After each stroke is drawn, rudimentary processing is performed on the stroke, reducing the stroke to an ellipse or a series of line and curve segments. A collection of spatially and temporally close strokes is chosen, and the line segments contained in the collection of strokes are then recognized as either an editing command or a viewable object.

            During the recognition stage, all stroke collections are examined to see if a particular stroke collection could be interpreted as a viewable object or an editing command. During the identification stage, a final interpretation is chosen, and a collection of strokes is identified as a viewable object or an editing command. All possible interpretations found in the recognition stage from the stroke collections are presented to the identification stage. The identification stage selects the final interpretation based on the following rules. 

DISCUSSION

            Tahuti combines the sketching freedom provided by paper sketches and the processing power available in an interpreted diagram. The system is based on a multi-layer recognition framework and recognizes objects by their geometrical properties, rather than requiring that the user draw the objects in a pre-defined manner. This system considers only groups of strokes that fall within a spatial bound. This spatial bound on its own may not be enough, especially for regions that contain many overlapping strokes. For example, if there are ten lines that all fall within a small region, to identify an arrow, the system still may have to try combinations of these lines.

Reading #29: Scratch Input Creating Large, Inexpensive, Unpowered and Mobile Finger Input Surfaces

COMMENTS ON OTHERS:

           Jonathan 

SUMMARY

              In this paper, the authors provide a new input technique that allows small devices to appropriate existing, large, passive surfaces such as desks and walls, for use as a kind of input device. This Scratch Input technique operates by listening to the sound of “scratching” (e.g., with a fingernail) that is transmitted through the surface material. This signal can be used to recognize a vocabulary of gestures carried out by the user. The proposed sensor is simple and inexpensive, and can be easily incorporated into mobile devices, enabling them to appropriate whatever solid surface they happen to be resting on. Alternately, it can be very easily deployed, for example, to make existing walls or furniture input-capable.

            To capture sound transmission through solid materials, authors proposed to use a modified stethoscope. This is particularly well suited to both amplifying sound and detecting high frequency noises. This is attached to a generic microphone, which converts the sound into an electrical signal. In this particular implementation, the signal is amplified and connected to a computer through the audio-input jack. Scratch Input’s non-spatial property gives it a significantly different character from many other surface input techniques and does preclude some uses. Results indicate participants were able to achieve an average accuracy of 89.5%. As hypothesized, accuracy suffered as gesture complexity grew. Gestures with two of fewer motions achieved accuracies in excess of 90%. 

DISCUSSION

            The Scratch Input, an acoustic-based finger input technique that can be used to create large, inexpensive and mobile finger input surfaces. This can allow mobile devices to appropriate surfaces on which they rest for gestural input. This revealed that Scratch Input is both easy to use and accurate on a variety of surfaces. Foremost, most mechanical sensors are engineered to provide relatively flat response curves over the range of frequencies that is relevant to signal. This is a desirable property for most applications where a faithful representation of an input signal – uncolored by the properties of the transducer – is desired. However, because only a specific set of frequencies is conducted through the arm in response to tap input, a flat response curve leads to the capture of irrelevant frequencies and thus to a high signal-to-noise ratio.

Reading #27: K-sketch: A 'Kinetic' Sketch Pad for Novice Animators

COMMENTS ON OTHERS:

            Francisco

SUMMARY

              The authors proposed an informal, 2D animation system called KSketch, the “Kinetic” Sketch Pad. K-Sketch is a pen-based system that relies on users’ intuitive sense of space and time while still supporting a wide range of uses. K-Sketch animations are often rough, but they are still useful in informal situations and as prototypes of formal animations. The goal of this project has not been to design novel interaction techniques but rather to focus on high-level choices about tool features. Thus, the authors conducted field studies to find out how an informal animation tool might be used and whether or not it could be made general-purpose.

From these interviews with nineteen animators and would be animators, authors compiled a library of 72 usage scenarios for an animation system. these results in more detail and describe a novel optimization technique that enable to make K-Sketch’s interface simultaneously fast, simple, and powerful.

            Process begins by reviewing interviews with animators and with non-animators. This is followed by an analysis of the library of usage scenarios collected and a description of interface optimization technique. Since many novice animators wish to do what experienced animators do, the authors began field studies by interviewing eight experienced animators to see how an informal tool would fit in their work process. K-Sketch currently supports all ten desired animation operations: Translate, Scale, Rotate, Set Timing, Move Relative, Appear, Disappear, Trace, Copy Motion, and Orient to Path. 

DISCUSSION

            These results show that K-Sketch’s simple interface has strong benefits. The simplicity of K-Sketch’s interface also meant less practice time was needed before tasks could be performed. These tools allow designers to build prototypes or storyboards of dynamic systems by creating sketches according to conventional visual languages.

Reading #26: Picturephone: A Game for Sketch Data Capture


COMMENTS ON OTHERS:

            Francisco 

SUMMARY

             This paper proposes a multi-player sketching game called Picturephone. Its purpose is to capture hand-drawn sketches and player-provided descriptions which can be used by other researchers to develop or test sketch recognition systems. Picturephone is not a new recognition system—it is a tool for capturing hand-made drawings in many domains by many people, along with human-classified descriptions. Picturephone is inspired by the children’s game called Telephone. In Telephone, a player privately describes something to the person to the left. That person then conveys the message to the person to their left, and so on. Over time the message may change drastically. Picturephone uses a web oriented client/server architecture and is known to run on Windows, Mac OS X, and Ubuntu Linux. Both client and server are written in Java. Communication is done with the standard HTTP protocol using the host web browser’s network connection, allowing the game to work unimpeded by firewall or router restrictions.

            There are three primary game modes: draw, describe, and rate. Players are randomly assigned one of these modes. In Draw mode (Figure 2), players are given a text description and are asked to draw it using the sketching surface at the right. A time limit is enforced to encourage simplicity. 

DISCUSSION

            Picturephone is the first instance of a class of planned sketching games that could provide researchers with a method to acquire data about drawings. This includes the physical act of sketching as well as how people describe those drawn elements. The paper is lack of most of its implementation details, and significantly overlaps with the paper described in Reading #24. Here it asks the player to draw using the description “three concentric circles”. After completing the drawing, the player hits ‘Done’. The game chooses among your preferred modes: sketching, describing, or rating.

The user gets a few ‘rate’ phases in a row here. It simply asks the player to rate how closely the two pictures match. The picture on the left was the basis for a description, and the one on the right was a sketch made based on that description. When a player rates these two drawings, points are assigned to the people who made both sketches as well as whoever made the mediating description. As you can see, the task of rating is rather subjective, but since it collects lots of ratings for the same pair it seems to work out fairly well. Eventually the game picks ‘describe’ mode, so here you see the player typing in a description about a floor plan layout of a square house with a bathroom and kitchen in the corners. This process continues until the player chooses to end their session.

Reading #25: A Descriptor for Large Scale Image Retrieval Based on Sketched Feature Lines

COMMENTS ON OTHERS:

            Jonathan

SUMMARY

              The main contribution of this work is a sketch-based query system for image databases containing millions of images. As most current retrieval algorithms for large image databases it is based on a small descriptor that captures essential properties of the images. A main feature is that it elegantly addresses the asymmetry between the binary user sketch on the one hand and the full color image on the other hand. The proposed descriptor is constructed in such a way that both the full color image and the sketch undergo exactly the same preprocessing steps to compute the descriptor. The resulting sketch based image retrieval system can be used by any novice user to quickly query the image database. The power of the system stems from exploiting the vast amount of existing images, which offsets obvious deficits in image descriptors and search.

The authors state that the descriptor’s performance is superior to a variant of the MPEG-7 edge histogram descriptor in a quantitative evaluation for which measured retrieval ranks of 27 sketches created from reference images out of the image database. 

DISCUSSION

            I’m totally perplexed. Is this related to sketch recognition? 

Reading #24: Games for Sketch Data Collection

COMMENTS ON OTHERS:

            Hong-Hoe (Ayden) Kim

SUMMARY

              This paper presents a multi-player sketching games to capture a data corpus of hand-drawn sketches and player-provided descriptions from many users on a wide range of subjects. Two systems with distinct game mechanics are described: Picturephone and Stellasketch. Picturephone has three primary game modes: draw, describe, and rate. Players are randomly assigned one of these modes. Stellasketch is a synchronous, multi-player sketching game similar to the parlor game Pictionary. One player is asked to make a drawing based on a secret clue. The other players see the drawing unfold as it is made and privately label the drawing. While Picturephone’s descriptions are meant to be used to recreate a drawing, Stellasketch’s labels simply state what the sketch depicts. Labels are timestamped, so they can be associated with sketches at various stages of completion.

The characteristics of the two games’ data differ. While Picturephone’s sketches are complete at the time when others describe them, a Stellasketch drawing is labeled as it is made. Furthermore, Picturephone descriptions are generally longer and in approximately complete sentences, but Stellasketch labels are often short noun-phrases. Because a Stellasketch drawing is labeled as it is made, players usually furnish multiple interpretations, and there is often significant agreement among players. Agreement indicates those interpretations are more ‘correct’. Sometimes labels cluster into more than one group. While Picturephone supports people to play at their own rate, a game of Stellasketch requires several people to play at the same rate. 

DISCUSSION

            This paper has presented Picturephone and Stellasketch, two sketching games for collecting data about how people make and describe hand-made drawings.

Reading #23: InkSeine: In Situ Search for Active Note Taking

COMMENTS ON OTHERS:

         Jonathan  

SUMMARY

            InkSeine is a TabletPC application that offers rapid, minimally distracting interactions for users to seek, gather, and manipulate the “task detritus” of electronic work (links to documents, clippings from web pages, or key emails on a topic) across multiple notebook pages. Search offers a facile means to assemble such collages of notes, documents, and bitmaps while keeping the user engrossed in the inking experience as much as possible. InkSeine’s primary work surface is an electronic notebook that allows users to jot ink notes on a series of pages. Thus, all of the search facilities that are the focus of this paper are designed in support of the inking task itself.

InkSeine provides a quick way to collect and annotate content from multiple documents. While current features for gathering content provide ways to drag out information from the search panel, there are opportunities to further leverage the value of in situ search by allowing the user to pull in material for searches from the notes.  InkSeine’s in situ ink search strategy helps to reduce the cognitive barrier between having the thought to do a search while inking, to actually capturing that thought, and potentially acting on it at a later time. 

DISCUSSION

            InkSeine is a Tablet PC search application that allows users to store a pointer to a search via a breadcrumb object intermixed with their handwritten notes. Hinkley discovered that conventional GUI tooltips could be easily blocked by the hand. InkSeine presented a variation of a theme in which gestures were shown in situ as highlighter annotations over application widgets; the annotations could be toggled on and off with a button press. This technique was well suited toward disclosing simple gestures associated with explicit UI widgets. However, with only a few gestures, the technique cluttered the workspace but did not provide support for accessing more detailed information about subtle or complex gestures or for displaying gestures that required a document context (e.g., a selection lasso).

Reading #22: Plushie: An Interactive Design System for Plush Toys

COMMENTS ON OTHERS:

            Hong-Hoe (Ayden) Kim

SUMMARY

            Plushie, is an interactive system that allows nonprofessional users to design their own original plush toys. To design a plush toy, one needs to construct an appropriate two-dimensional (2D) pattern. However, it is difficult for non-professional users to appropriately design a 2D pattern. Plushie, allows the user to create a 3D plush toy model from scratch by simply drawing its desired silhouette. The user can also edit the model, such as by cutting the model and adding a part, using simple sketching interface. The resulting model is always associated with a 2D pattern and the 3D model is the result of a physical simulation that mimics the inflation effect caused by stuffing.

The user interactively draws free-form strokes on the canvas as gestures and the system performs corresponding operations. The system also provides some special editing operations tailored for plush toy design including create, cut, create parts, pull, insert and delete.  The authors used a standard triangle mesh for the representation of 3D model and the 2D patches with relatively coarse mesh (1000-2000 vertices) to achieve interactive performance. Each vertex, edge, and face of the 3D mesh is associated with corresponding entities in the 2D mesh. A 3D mesh is always given as a result of applying a physical simulation to the assembled 2D pattern. To be more precise, the physical simulation applied to the 3D mesh is governed by the rest length of each edge, which is defined in the 2D mesh geometry.  

DISCUSSION

            Creating 3D models is often a hard and laborious task due to the complexity and diversity of shapes involved, the intricate relationships between them, and the variety of surface representations. Current high-end modeling systems such as Maya, AutoCAD, and CATIA incorporate powerful tools for accurate and detailed geometric model construction and manipulation. These systems typically employ the WIMP (Window, Icon, Menu, Pointer) interface paradigm, which are based on selecting operations from menus and floating palettes, entering parameters in dialog boxes, and moving control points.

In this case, sketched input is used to define an initial state of a complex physical simulation or procedural model, domains that are typically encumbered with many parameters and initial settings to define. Mori and Igarashi provide an intriguing example of how SBIM techniques could be integrated with physical simulation: “if one can run an aerodynamic simulation during the interactive design of an airplane model, it might be helpful to intelligently adjust the entire geometry in response to the user’s simple deformation operations so that it can actually fly.” Exploring the output space of a procedural or physical model can be much more natural and efficient with free-form gestures, a notion that needs to be explored more fully in the future.

Reading #21: Teddy: A Sketching Interface for 3D Freeform Design

COMMENTS ON OTHERS:

            Francisco

SUMMARY

            The paper proposes the “Teddy”, an idea of creating a sketching interface for designing 3D freeform objects. This allows the user to draw 2D freeform strokes interactively specifying the silhouette of an object and the system is capable of automatically constructing the 3D polygonal surface model based on the strokes.

The user does not have to manipulate control points or combine complicated editing operations. Using this technique, even first-time users can create simple, yet expressive 3D models within minutes. 

Teddy’s physical user interface is based upon traditional 2D input devices such as a standard mouse or tablet. As soon as the user finishes drawing the stroke, the system automatically constructs a corresponding 3D shape. The program supports operations of creating new object, painting and erasing on the surface, Extrusion, cutting, smoothing and transformation. In order to remove noise in the handwriting input stroke and to construct a regular polygonal mesh, every input stroke is re-sampled to form a smooth polyline with uniform edge length before further processing 

DISCUSSION

            The authors follow the general philosophy of keeping the user interface simple by inferring the intention of a few, easy-to-learn commands, rather than providing an exhaustive set of commands and asking the user to set several parameters for each one. However, this is done by limiting the complexity and types of shapes that can be created by the user. Furthermore, the proposed system is not provided with a resultant accuracy measures and comparison parameters.

Wednesday, December 08, 2010

Reading #20: MathPad2: A System for the Creation and Exploration of Mathematical Sketches

COMMENTS ON OTHERS:

            Francisco 

SUMMARY

              This paper presents MathPad2, a prototype application for creating mathematical sketches.  MathPad2 incorporates a novel gestural interaction paradigm that lets users modelessly create handwritten mathematical expressions using familiar math notation and free-form diagrams, including associations between then two, with only a stylus.

According to authors, Mathematical sketching is the process of making simple illustrations from a combination of handwritten 2D mathematical expressions and sketched diagrams. Combining mathematical expressions with diagram elements, called an association, is done either implicitly using diagram labels as input to an inferencing engine or manually using a simple gestural user interface. 

DISCUSSION

            It is hard to understand the novelty of the work other than using a gesture recognition system applied towards identifying mathematical expressions. Most of the gestures used for the work are of hackneyed nature, and the use of lasso capability to group stands as something original. In my personal opinion, the work is more of a functionality of a regular character recognition system and gestures are just used to trigger an execution of a function processing, or combining, or editing/deleting.

Reading #19: Diagram Structure Recognition by Bayesian Conditional Random Fields

COMMENTS ON OTHERS:

            Jonathan  

SUMMARY

              The paper proposes recognition of hand-drawn diagram structures using Bayesian conditional random fields (BCRFs). BCRFs are a generalization of Bayes point machines or discriminative Bayesian classifiers for single elements. The method jointly analyzes all drawing elements in order to incorporate contextual cues. According to authors, the advantage of BCRFs over conventionally-trained CRFs include, model averaging and automatic hyper parameter tuning.

            The application of BCRFs to ink classification is to discriminate between the containers and connectors in a drawing of organization charts. This particular task includes, subdivision of pen strokes into fragments, construction of conditional random field on fragments, BCRF training and inference on network. In the first step, the strokes are divided into simpler components called fragments. Fragments should be small enough to belong to a single container or connector. In contrast, strokes occasionally span more than one part when drawn without lifting a pen. The fragments are selected as strokes that form a straight line segments. In the process of conditional random field on the fragments, each fragment is represented by a node in the network. 

DISCUSSION

            The proposed work shows an interesting method to segment ink features from a sketch application. The method is mathematically rigorous and therefore hard to understand by the content. Great deal of information of the feature extraction omitted from the paper, and therefore makes the works shine a bit less. Also note that the BCRFs accuracy is less compared to the joint performance of BCRFs-ARD.