|09:30-10:30||Inauguration of Workshop|
|12:30-13:30||Technical Talks : Modeling & Representing Tangible Heritage of Hampi
|14:15-15:00||Technical Talks : Modeling and Representing Tangible Heritage of Hampi (Contd.)
|15:15-16:15||Technical Talks : Analysis and Digital Restoration of Artefacts
|18:00-19:00||Inauguration of Exhibition (Talk on Exhibits and Hampi)|
Technical Talks : Analysis and Digital Restoration of Artefacts
Technical Talks : Social Life, Living traditions
Technical Talks : Ontology and Knowledge Bank
|14:40-15:00||Invited talk: Big Data, cloud and e-Heritage by Dr. Hiranmay Ghosh, TCS|
|15:00-15:20||Invited Talk : Archival photos of Hampi and Digital approaches to photographic preservation by Ms. Shilpi Goswami, Alkazi Foundation|
The Age of Experience examines new paradigms for developing embodied museum experiences in the 'age of information'. Using datasets representing intangible and tangible heritage, the talk explores interactive applications inside a series of nine fully immersive visualization systems.
Imaging data at world heritage sites is fundamental to archaeological, conservation and preservation processes. And although this high-resolution data is accumulating daily in the archives of practitioners little of it finds its way into the public domain. Large-scale display systems offer the opportunity for this data to be experienced at one-to-one-scale in the form of augmented digital facsimiles. Such an approach has a significant impact on visitors' cognition and also provides a powerful alternative for giving access to heritage sites under threat from tourism, conflict and climate change. At the same time, visitors to museums are increasingly interested in intuitively exploring and re-using the large digital datasets created through heritage documentation. This desire for creative engagement poses significant experimental and theoretical challenges for the museums sector. The Age of Experience engages cultural heritage data in contemporary museum discourses and future interpretation practices.
The museum installations described in this lecture include the world heritage sites of Angkor in Cambodia, Dunhuang in Gansu Province China, the Monuments at Hampi in South India and numerous sites throughout Turkey. Other works emphasize the visualization of cultural collections ranging from the 70,000 records from the web-based WW1 archives of Europeana (the world's largest online cultural collection) to 100,000 objects from the collections of Museum Victoria, Melbourne, in the world's largest data browser. Two works based on Pacifying of the South China Sea Pirates' scroll painting at Maritime Museum, Hong Kong will also be described. New projects under development include: 3D documentation of the living archives of South Chinese Kung Fu; the 360-omnidirectional installation Atlas of Maritime Buddhism which tells the unique story of the spread of Buddhism from India to Korea through the seaports of South East Asia from the 2nd C. BC to the 12th C. AD. and; a fulldome installation planned for the Chhatrapati Shivaji Maharaj Vastu Sangrahalaya, Mumbai.
The main objective of the project is to create algorithms and techniques to acquire a three dimensional digital replica of complex structures spread over a large area. The techniques developed are applied to Hampi, a world heritage site. In addition to acquiring the geometry and surface properties, we also research efficient representation and visualization of this data and provide tools and methods for users to experience the captured models, to virtual walkthrough and explore the digital recreations.
For the acquisition we rely on multi modal input using technologies like laser scanners, color cameras, and depth sensors. We align and fuse geometric constructions from different modalities through a step of registration. We have extended structure from motion (SfM) a state-of-art approach for multi-view 3D reconstruction from images and developed techniques for large scale (relatively) sparse geometric constructions and simultaneously dense reconstructions of smaller parts. We also provide ability to generate high resolution point cloud from the point cloud obtained from depth camera Kinect by using additional high definition cameras.
In addition to creating geometric models, we investigate computational modeling of restoration techniques, and artistic rendering of heritage artifacts. We also explore efficient visualization of large models with augmented reality and user-experience authoring.
Hampi has been chosen as a test-bench for developing our techniques. Within Hampi, we concentrate on Vittala Temple Complex, and demonstrate our techniques on it. The project has greatly benefitted from the collaboration from other partner institutes especially BVBCET, NID, NIAS, IIT Bombay, IIT Madras and IIIT Hyderabad.
The importance of digitization of heritage data cannot be underestimated. Given the vulnerability of heritage sites due to natural calamity and pressure of urbanisation, it is imperative to develop methods that shall aid in the process of preservation for posterity. While credible initiatives have been launched in several countries to restore damaged heritage, the conservation architects at such organizations would need a reference to perform the task of restoration.
While it is important to digitize heritage sites `as is', building 3D models of damaged archaeological structures can be visually unpleasant due to presence of large missing regions. In this paper, we address geometric and photometric reconstruction of such naturally occurring large damaged regions (or holes) in 3D digital models. Without constraining the size or complexity of the damaged region, the missing 3D geometry inference problem is solved by making use of geometric prior from self-similar structures which provide a salient cue about the missing surface characteristics that may be unique to an object class. The underlying surface is then recovered by adaptively propagating 3D surface smoothness from local geometric information around the boundary of the hole. This process exploits the cue provided by the available self-similar examples. We propose a methodology to effectively harness the geometric prior by employing a tensor voting-based method when multiple self-similar examples are available. For photometric completion of the geometrically in-painted region, a low-rank, sparse formulation is discussed. The proposed geometric and photometric in-painting pipeline allows for the creation of visually pleasing 3D models of archaeological structures, which can then readily serve as a natural addition to heritage visualization, not merely for the purpose of preservation but also for applications such as a `virtual tour'. The performance of the proposed methods on holes with different complexities and sizes will be shown on synthetic as well as real data.
In this work, we develop efficient and effective methods for recovering the 3D structure of large scale objects such as heritage monuments. There are several challenges that arises both from the scale of data as well as the expectations of the quality of results as compared to more mundane objects. The collection of images for large scale monuments itself is a challenge to structure from motion (SfM) techniques as the process of acquiring multi-view data for every small structure in a monument would be prohibitively expensive. We look at solutions to both of these problems. Heritage monuments are often covered with relief carvings and they present an huge challenge to existing 3D recovery methods. The difficulty is due to the presence of repetitions resulting in ambiguity of matches as well as the requirement of large number of images due to the intricate nature of these carvings.
To overcome this, we present a method for recovering the 3D structure of a relief carving from a single image. This is an ill posed problem and has been attempted before with limited success. We employ the additional knowledge of the data being relief carvings to develop a learning and estimation method for such structures. The prior knowledge is acquired from high quality image-depth pairs of relief structures. A coupled sparse dictionary is learned from the training data, which is used to estimate the depth maps of new relief images. The resulting depth map is quite representative of the overall structure, but lacks finer details. We combine the results with depth from shading, which is detailed but noisy, to obtain high quality reconstructions.
To deal with the prohibitive computing requirements of large-scale matching, we have developed a new multistage approach for SfM reconstruction of a single component. Our method begins with building a coarse 3D reconstruction using high- scale features of given images. This step uses only a fraction of features and is fast. We enrich the model in stages by localizing the remaining images to it and matching and triangulating remaining features. Unlike traditional incremental SfM, localization and triangulation steps in our approach are made efficient and embarrassingly parallel using geometry of the coarse model. The coarse model allows us to use 3D-2D correspondences based direct localization techniques to register remaining images. We further utilize the geometry of the coarse model to reduce the pair-wise image matching effort as well as to perform fast guided feature matching for majority of features. Our method produces similar quality models as compared to incremental SfM methods while being notably fast and parallel. The algorithm can reconstruct a 1000 images dataset in 15 hours using a single core, in about 2 hours using 8 cores and in a few minutes by utilizing full parallelism of about 200 cores.
In this project we discuss the framework for a realistic walk-through of cultural heritage sites. The framework includes 3D data acquisition, different data processing steps, coarse to fine reconstruction and rendering to generate realistic walkthrough. We propose a coarse to fine 3D reconstruction of heritage sites using different 3D data acquisition techniques. The data acquisition step includes acquisition of 3D data for different modalities like CAD model, Single-view model, Kinect model and Multi-view model. We develop geometry based data processing algorithms using Riemannian metric tensor and Christoffel symbols as a novel set of features. 3D object categorization, decomposition, super resolution and hole-filling are performed to refine the point cloud data in the data processing step. In the rendering step, the refined data is fed to coarse to fine 3D reconstruction stage. We generate a walkthrough of the cultural heritage sites using the coarse to fine 3D reconstructed models and demonstrate the proposed framework using a walkthrough generated for the Vittala Temple at Hampi.
We model 3D object as a piecewise smooth Riemannian manifold and propose metric tensor and Christoffel symbols as a novel set of features for 3D data processing. The categorization of 3D objects is carried out using polynomial kernel SVM classifier by capturing the global geometry using the combination of metric tensor and Christoffel symbols. The point-cloud data obtained from the low-resolution 3D scanner like the Microsoft Kinect or from sparse reconstruction algorithms fail to portray all the details in a model's surface resulting in a low-resolution point-cloud data. A Kernel based SVM learning framework is employed to decompose a given 3D model into basics shapes viz., sphere, cone and cylinder using metric tensor and Christoffel symbols as a set of novel geometric features. The decomposed models are then independently super-resolved using selective interpolation techniques. The independently super resolved algorithms are merged to obtain the final super resolved model. The 3D data acquired consists of missing regions or holes due to occlusions in the surface to be scanned. The boundary detection algorithm is used to identify holes and the neighborhood of the hole is decomposed into basic shapes using a kernel based SVM learning framework with metric tensor and Christoffel symbols as features. The centroid of the hole region is computed and the selective surface interpolation is carried out along the directional vector. The coarse to fine level 3D reconstruction is achieved by registering the coarse level 3D models with the fine level 3D models. The fine level 3D models are superimposed on the coarse level 3D models by interactively selecting the correspondence points in the model and registered using Iterative Closest Point (ICP). The rendering of the reconstructed models is carried out using either a rendering engine like OGRE 3D or a gaming engine like Unity 3D.
Indian temple architecture gives us a rare insight into design, construction, proportion and scale. Dravidian temples, especially Vijayanagara architecture and in particular Hampi architecture is an impeccable synergy between structural innovation and architectural expression. The 'Vittala temple complex' at Hampi, Karnataka, India has been taken for study, documentation, analysis of design elements and 3D virtual modeling /reconstruction of the complete temple complex. The project is an insight of graphical, pictorial and digital reconstruction using Auto- CAD, Google Sketch-up and Kinect software. The existing temple complex was extensively studied, analyzed and documented. A 3D model was built. The scale and proportion of the depth of the plinth, column, roof, parapet and complex as a whole is studied. The study of the structural system and architectural grammar of the existing elements led to the understanding of missing elements. The missing part of the temple example the shikhara over the chariot has been reconstructed with the help of images of the past, studying the proportions and applying principles of architecture and also the remains of existing parapet in the mandapa/pavilion are studied thoroughly and replicated using this tool. The tool has been used to facilitate the visual re-construction to achieve the architecture of the temple in its original form.
We propose a new technique for building a system for multi-modal rendering of 3-D cultural heritage objects represented as a dense depth map data. The primary emphasis here is on how to include the kinesthetic sensation of touching an object. Rendering of surface properties like texture and friction is found to enrich the user's experience in the virtual world. We include scalability, rotation, translation and stereoscopic display of 3- D models as additional features to enhance the realism in experience. We integrate audio rendering with respect to a spatially segmented 3-D object. We demonstrate this with respect to the musical pillars at Hampi. The pillars in this temple have musical columns which produces distinct sounds when struck. We also conducted a user survey on a few subjects and observed that hapto-visual and auditory rendering of virtual 3-D models using the proposed method greatly augmented the user's experience.
For a combined hapto-visual rendering, the 3-D object surface is displayed as a simple quad mesh formed out of the depth values. We have opted for the mesh- based graphical display in order to give a better perception to the viewer since using point cloud data for graphic display would result in gaps in the visually rendered image. We have used the stereoscopic display technique for creating the effect of depth in the image by presenting two offset images on the screen corresponding to two different points of projection. Anaglyphic glasses can be used to view such displays. Sounds are incorporated in the rendering framework by playing appropriate audio files based on the position of haptic probe inside the virtual environment. During haptic rendering whenever a collision takes place for the first time with a specific pillar, the corresponding note is played, the volume of which is made proportional to the rendered force.
Vijayanagara period sculptures in stone and bronze represent distinctive and rich stylistic traditions. This paper explores the usefulness of bringing together aspects of digital rendition together with iconographic and iconometric studies to better document and enhance the understanding of Vijayanagara sculpture. This is relevant especially given that the images have had their own individual histories over the course of time, as for example some of the images that were damaged or broken. Studies on iconometric conventions and iconographic aspects have a role to play in terms of gaining more insights into the traditional modelling of such sculptural examples. One of the points of interest in terms of studies in 3-D modelling and digital restoration is that apart from in situ examples of stone sculpture extant in the numerous monuments at the World Heritage site of Hampi, there are several examples of sculpture, to be found in collections such as Kamalapura museum, Hampi, and from the region of Hampi, some which have missing body parts and which include for example stone portrait sculptures. Another aspect that the study touches upon is to identify similar themes executed in stone and bronze and for purpose of comparisons between the iconometric aspects of modelling and portrayals of themes in different sculptural media. Thus for example, a bronze Lakshmi Narasimha image has also been studied with this in mind in relation to the well known stone Ugra Narasimha which in itself represents a reconstructed version of a damaged image. A 3-D laser scanned digital image generated of the Ugra Narasimha image from Hampi undertaken by KCST has been examined in a preliminary fashion in terms of iconometric and iconographic aspects. Vijayanagara era sculptures and bronzes in the collection of Chandragiri Museum in the Chandragiri fort were also explored as a little known collection of this era and since some of the sculptures were originally found in Hampi area. The classificatory aspects have also been explored in relation to the ontological approaches being explored by the IDH group from IIT Delhi. Thus the paper points to some significant directions and inputs that could contribute to the development of 3-D modeling and exploring digital renditions of sculpture.
Restoration of heritage artifacts such as murals and paintings of the past is an important task for preservation of social, cultural and political history of a nation. As, being in the temples in India, a significant share of murals are not accessible for physical restoration. This motivates many researchers to put effort to restore such priceless paintings and reliefs digitally. In this work, we propose an interactive algorithm for coherent image completion that would be useful for the paintings containing repetitive patterns under constrained environment where user needs to mark the target (damaged) region together with the source (intact) texture region. To remove spurious noises as well as to preserve edges, an edge enhancing diffusion scheme is used. The said diffusion technique is based on Beltrami-kernel flow and a novel patch-based high frequency enhancing method applied in alternating sequence.
Establishing the correspondence of features is a longstanding problem in Computer Vision. It is an indispensable module for many tasks including Image Stitching, Disparity Map Estimation, Structure from Motion, 3D-Reconstruction, Object Tracking and Object Identification.
Given two sets of features (or keypoints) extracted from two images, the task is to match the features from one to those in the other such that there is a visual correspondence in the matching. To achieve this, the regions around the keypoints are analyzed for visual characteristics and represented using feature or region descriptors. Feature descriptors are usually vectors or histograms that bring out properties such as the distribution of intensities, gradients, edge patterns or texture that can be matched using suitable distance metrics. There are many feature descriptors that have been proposed in the literature, many of which are appropriate for specific kinds of scenes.
Order-based descriptors are constructed by comparing the pixels in a region with their neighbors. They can be appropriately designed to be made invariant to changes in illumination, especially those in under-saturation and over-saturation regions, pixel noise, and robust to blur and compression artifacts. Along with these, they can also be built to portray local edge patterns that bring out the shape characteristics of a region. In this chapter, we study the applicability of order-based descriptors for matching keypoints extracted from architectural scenes.
Scenes of architecture and monuments are characterized by regions that are textured and varying in depth. Also, many of the images of these scenes are shot with sufficient area of homogeneous regions(sky as the background or a plain landscape in front) under varying light conditions. This work studies the effect of such challenges on the performance of these descriptors for matching features in 2 applications - image stitching and 3D-reconstruction (using tools like Bundler and Visual SfM) and aims to understand the scope of applicability of these descriptors.
This presentation is a short overview of local communities in Hampi and the role they played as custodians of Hampi Heritage. Talk will touch upon customs and rituals that were in practice several centuries ago and still practiced by local communities. Importance of their role in preserving Hampi heritage for future generations.
Historical monuments are considered as one of the key aspects of one's cultural heritage. Unfortunately, due to a variety of factors the monuments get damaged. For images of historic monuments in particular, there is a consensus to fill-up the cracks so that one can view these in their undamaged form. Thus, we are not talking about image restoration, but about object completion by digitally repairing cracks / damaged regions that the physical objects have. In this talk, we discuss a technique for automatically detecting the cracks in photographs of monuments. Unlike the usual practice of manually selecting the mask for inpainting, the detected regions are supplied to inpainting algorithms. Thus, the process of digital repair using inpainting is completely automated. In addition we discuss two new image inpainting methods. Experiments have been carried out on videos captured from the heritage site at Hampi, India. The proposed image inpainting methods are based on a) minimizing difference in curvature and b) patch matching in a multi-resolution framework.
Recent approaches on single image super-resolution (SR) have attempted to exploit self-similarity to avoid the use of multiple images. In this talk, we first discuss a SR method based on self-learning and Gabor prior. Given a low resolution (LR) test image I0 and its coarser resolution version I-1, both captured from the same camera, we first estimate the degradation between LR and HR (I1) images by constructing the LR-HR patches from LR test image, I0. The HR patches are obtained from I0 by searching for similar patches of the same size in I-1. A nearest neighbor search is used to find the best LR match which is then used to obtain the parent HR patch from I0. All such LR-HR patches form the self-learned dictionaries. The HR patches that do not find LR match in I-1 are estimated using self-learned dictionaries constructed from the already found LR-HR patches. A compressive sensing-based framework is used to obtain the missing HR patches. The estimated LR-HR pairs are used in LR image formation model to compute the degradation. Finally, a regularization based approach in which a new prior, called Gabor Prior based on the outputs of a Gabor filter bank is used to obtain final SR. The experimental results show the effectiveness of the proposed approach.
A significant amount of research has been carried out in the direction of reading inscriptions from monuments around the world. Several methods have been proposed for detection of text, localization and extraction of text from images of inscriptions. But, the problem of text extraction becomes extremely difficult when the difference in the text (foreground) and the background is very marginal, the background is textured, or the background and foreground are similar. Such is the case of camera-held images of inscriptions at the sites of historical monuments. Figure 1 shows an image of inscription found in world heritage site "Hampi". These inscriptions are generally found engraved into/projected out from, stone, or other durable materials. However, due to effects of uncontrolled illuminations, wrapping, multilingual text, minimal difference between foreground and background images, and the distortion due to perspective projection as well as the complexity of image background, extracting text from these images is a challenging problem. In this paper we discuss a new technique for text extraction using variants of blind source separation algorithms.
In this work, we design and demonstrate a complete pipeline for multimedia retrieval on a mobile device with applications in Digital heritage. We target the use case of a tourist or a visitor at a heritage site, who wishes guide herself by clicking an image of an interesting structure to get information about the same. This requires efficient mobile based instance retrieval techniques over a fairly large dataset of images. Users query the system with an example image captured with a mobile device and obtain the related information on the mobile device instantly. More over, the image specific annotations (descriptions of the regions or objects present in the image) are registered and overlaid on the query image. Our objective is to carry out the computations on the mobile device itself.An effective visual search task on mobile device requires a significant reduction in the visual index size. To achieve this, we describe a set of strategies that can reduce the visual index to be compatible with the low to mid end mobile phones. While our proposed reduction steps mildly affect the overall mean Average Precision, they are able to maintain a good Precision for the top K results. We argue that for such offline application, maintaining a good precision at the top of the list is sufficient.
Such an instance retrieval framework depends on a well annotated dataset of images to retrieve from. Photos from tourist and heritage sites can often be described with detailed and part-wise annotations. Manually annotating a large community photo collection is a costly and redundant process as similar images share the same annotations. Hence, we also present an interactive web-based annotation framework that allows multiple users to add, view, edit and suggest rich annotations for images in community photo collections. Since, distinct annotations could be few, we have an easy and efficient batch annotation approach using an image similarity graph, pre-computed with instance retrieval and matching. This helps in seamlessly propagating annotations of the same objects or similar images across the entire dataset.
Effective and efficient computer vision modules on mobile devices can open up a wide spectrum of applications on the mobile devices. We also investigate the possibility of adapting augmented reality applications on mobile devices, and investigate a systematic method to adapt the data intensive computer vision modules to a compact form with minimal loss in performance.
This paper would look at Reconstructions of Bazaar Streets of Hampi. The reconstructions of Bazaar Streets has been taken under two main components:
1. Architectural Reconstruction of Bazaar Street
2. Reconstructions of Social life and Bazaar Scenarios - Clothing & Material Culture, Bazaar activities
At first, we shall discuss urban level context at city level to arrive at probable linkages to the various bazaar streets through roadways and three level hierarchical gateways. This would show the various architectural typologies of the gateways along a roadway and linkages to various bazaars. Next, we shall look at the method of preparing the street footprint and then using this footprint for mapping various structures for bazaar streets with Virupaksha and Vitthala Bazaar Street as examples. We shall explore architectural Reconstruction methodology as comparative analysis for the two streets and highlight the different approaches and the rationale for the same.
We shall discuss reconstructions of Social life and Bazaar Scenario. We focus on Clothing Style and Material Culture and would discuss literary resources that looked at Vijayanagara material culture, to arrive at an understanding of reference terms and vocabulary for clothing. We then studied and analysed murals at Rangamantapa, Virupaksha temple, Lepakshi Temple murals, Stucco work at gopurams of Virupaksha and Krishna, Relief work at Mahanavami Dibba to understand the clothing and garment styles of Vijayanagara period. We shall also outline the mapping of various street activities to various locations on Bazaar street to arrive at conjectures for Bazaar scenarios that shows people and their activity within an architectural setting and a zone of the street.
Virtual characters are used extensively in virtual, augmented and mixed reality digital heritage applications. Creating these believable characters requires a lot of effort. The characters have to be modeled, textured, rigged and animated.
We present a system that creates a mesh model of the character by deforming a template mesh to match a point cloud captured from a system of depth cameras. The process maintains the topology of the template mesh in the process thereby producing an avatar mesh of proper topology at the end of the process. We validate the shape of the mesh by comparing with actual anthropological measurements from the real user with measurements on the mesh. We also capture motion data from a single depth camera and use that to animate the created mesh. The entire process is mostly automatic and other than the depth cameras (Microsoft Kinect cameras), no specific hardware is required.
We then present extensions to available open-source animation authoring environment in Blender that allow us to synthesis character animation from pre-recorded motion data. Though the techniques for these are well known, we believe our method to be the first of its kind available in Blender. We then proceed towards enhancing the secondary animation aspects like animation of the character's garments. Physics-based simulation of virtual garments has been around in the computer graphics community for some time, however, animation of Indian garments present some unique challenges. We present an attempt toward solving some of these.
We also briefly discuss changes character response based on user position and gaze, for interactive characters.
The Story of Human evolution from biological perspective might still be a controversy. But the symbiosis of cultural components like song, music, drama, skills, cuisine, annual festivals, crafts, oral traditions through- out the evolution of human civilization and society is undebatable. Indian heritage has always given major importance to the celebration of festivals. Apart from spiritual and religious importance, the indian societies have incorporated social, cultural and political importance to the celebrations. In this paper, we shall examine the tradition of Dasara festival. We shall present a study of the diversity of celebration over a wide geographical area, that is from Hampi to Mysore.
The Tradition and History of Royal Dasara can be traced back to the Vijayanagara Empire[1336 A.D-1646 A.D]. Keen on protecting the Hindu religion and culture from destruction by Muslim invaders, rulers of the time laid great emphasis on the restoration of temples and religious festivals during their reign. During the period of Deva Raya or Krishna Deva Raya of the Vijayanagara empire, The Navaratri festival started receiving royal patronage, and it acquired the status of a state festival. It was celebrated with great pomp, enthusiasm and fervor, and attracted visitors from many part of the country and even abroad. Political, administrative, religious and social significance with the appreciation of the laity added to its grandeur. We can get a glimpse of the historical past by the documentation of foreign travellers to the Vijayanagara Kingdom during their reign, and witnessed the life style of the Great Kings and the locals.
In this study, we have proposed a new paradigm for heritage preservation - 'an Intellectual Journey', which is more advanced than physical explorations of heritage sites and virtual explorations of monuments and museums. This paradigm proposes an experiential expedition into a historical era by using an ontology to inter-link the digital heritage artefacts with their background knowledge. A multimedia ontology encoded in the Multimedia Web Ontology Language (MOWL) is used to illustrate this paradigm by correlating the digital artefacts with their history as well their living context in today's world. The user experience involves an ontology-based traversal of a heritage theme, with an ontology-guided navigation through space and time and a dynamic display of different kinds of media. The theme selected for this application is that of the Girija Kalyana - the marriage of Shiva and Parvati, with its context related to the World heritage site of Hampi, in the state of Karnataka, India.
This study also proposes cross-modal retrieval through an archiving scheme for heritage mural paintings. The mural paintings typically depict stories from folklore, mythology and history. These narratives provide content-based correlations between different pieces of art. Our e-heritage scheme for archiving the mural paintings is based on an ontology which captures the background knowledge of these narratives. Media features and patterns derived from the mural content are used to enrich the ontology with multimedia data. Besides the mural content and its knowledge, the ontology also helps encode other aspects of the mural paintings like their painting style, color, physical location, time-period, etc., which are important parameters of their preservation. We propose a framework to provide cross-modal semantic linkage between semantically annotated content of a repository of Indian mural paintings from Hampi, and a collection of labelled text documents of their narratives. This framework, based on a multimedia ontology of the domain, helps preserve the cultural heritage encoded in these artefacts.
The IDH-Hampi project has been running for approximately 4 years now with multiple institutions involved and a plethora of technology and cultural ideas in play. It has been an extraordinary multi-disciplinary effort and has generated a lot of knowledge in a "bottom up" or inductive fashion. The "knowledge bank" is tasked now with integration of the knowledge generated and with creating a formal structure that would serve as both a digital (networked) library for scholars and a virtual museum for the education and enjoyment of the general public.
We have proposed that the two user paradigms may actually require different implementations. While the former, the digital library, is a classical problem of a semantically interconnected network of distributed repositories of knowledge, the latter would require a more cloud-like infrastructure since it would require computational resources to be brought to data rather than the reverse. As these frameworks evolve it will provide reusable infrastructure and also more deductive perspectives on digital heritage projects to be pursued in the future.
The knowledge bank project is in its infancy and hence this presentation will present the technology framework ideas more in the abstract. The speaker will use examples from the domain of mural paintings as heritage artefacts to illustrate both the use of the digital library to address debates in the art history of Vijayanagara mural paintings and the experience of a virtual museum that allows for an interactive experience with the exquisite mural paintings from the temples of the Vijayanagara empire.
1 Prof S Settar, the Principal Investigator of this project and a heritage scholar of repute, was unfortunately unable to attend this symposium due to prior commitments. Hence this presentation, by the co-PI, is focused more on the technological aspects of the project.
Professor Sarah Kenderdine researches at the forefront of interactive and immersive experiences for museums and galleries. In widely exhibited installation works, she has amalgamated cultural heritage with new media art practice, especially in the realms of interactive cinema, augmented reality and embodied narrative. She is a pioneer in panoramic and stereoscopic display systems and content creation. Sarah concurrently holds the position of Professor, National Institute for Experimental Arts (NIEA), College of Fine Arts, University of New South Wales (2013-) and Special Projects, Museum Victoria, Australia (2003-). She is Adjunct Prof. and Director of Research at the Applied Laboratory for Interactive Visualization and Embodiment (ALiVE), City University of Hong Kong and Adjunct Prof. at RMIT. Recent books include Theorizing Digital Cultural Heritage: a critical discourse, Cambridge: MIT Press, 2007 and PLACE-Hampi: Inhabiting the Panoramic Imaginary of Vijayanagara, Heidelberg: Kehrer Verlag, 2013.
In the last 10 years Kenderdine has produced over 60 exhibitions and installations for museums worldwide and written 35 publications. In 2014, she was awarded the CHASS Prize for Distinctive Work. In 2013, the International Council of Museum Award (Australia), the Australian Arts in Asia Awards Innovation Award and the Tartessos Prize 2013 for contributions to virtual archaeology worldwide and, the Digital Heritage International Congress & IMéRA Foundation Fellowship (Aix-Marseille University).