Manish K Thakur

Sign language recognition using Microsoft Kinect

In last decade lot of efforts had been made by research community to create sign language recognition system which provide a medium of communication for differently-abled people and their machine translations help others having trouble in understanding such sign languages. Computer vision and machine learning can be collectively applied to create such systems. In this paper, we present a sign language recognition system which makes use of depth images that were captured using a Microsoft Kinect® camera. Using computer vision algorithms, we develop a characteristic depth and motion profile for each sign language gesture. The feature matrix thus generated was trained using a multi-class SVM classifier and the final results were compared with existing techniques. The dataset used is of sign language gestures for the digits 0-9.

A performance analysis of objective video quality metrics for digital video watermarking

Rapid growth in sharing information using internet technologies laid the threats of copyright protection for digital multimedia contents. One of the remedy for intellectual property protection is Watermarking. This paper deals with numerous available digital video watermarking techniques and raised the issue of quality and size of watermarked video. Through experimental study, it has been shown that available quality metrics like, Peak signal to noise ratio (PSNR) and Structural similarity index (SSIM) are incapable to deal efficiently to analyze the quality degradation but out of two it has been claimed that SSIM results are more accurate as it is closer to the results obtained during subjective quality analysis. Further, using available video watermarking tools, variation in size of video has been analyzed.

Architectural space planning using Genetic Algorithms

This paper presents a system which can find out space planning for a single flat, arrangement of several flats on a single floor and extend the design for each floor and find out collective plan for a multi-storey apartment building. At each level it generates a plan which supports quick evacuation in case of adversity. Starting with design specifications in terms of constraints over spaces, use of Genetic Algorithm leads to a complete set of consistent conceptual design solutions named topological solutions. These topological solutions which do not presume any precise definitive dimension correspond to the sketching step that an architect carries out from the design specifications on a preliminary design phase in architecture. Further, door placement algorithm has been proposed with modifications in existing Dijkstras algorithm and dimensions analysis is carried out for the designs selected by the user. If the user wishes to generate a plan for many floors, inputs are taken accordingly and plan is generated which is efficient in terms of evacuation.

Architectural layout planning using genetic algorithms

This paper presents a model which generates architectural layout for a single flat having regular shaped spaces; Bedroom, Bathroom, Kitchen, Balcony, Living and Dining Room. Using constraints at two levels; Topological (Adjacency, Compactness, Vaastu, and Open and Closed face constraints) and Dimensional (Length to Width ratio constraint), Genetic Algorithms have been used to generate the topological arrangement of spaces in the layout and further if required, feasibility have been dimensionally analyzed. Further easy evacuation form the selected layout in case of adversity has been proposed using Dijkstras Algorithm. Later the proposed model has been tested for efficiency using various test cases. This paper also presents a classification and categorization of various problems of space planning.

Intelligent Mail Box

In 21st century, email is one of the most effective ways of written communication due to its easy and quick access. But now days with each individual receiving large number of emails, mostly promotional and unnecessary mails, organization of emails in individual’s inbox is a tedious task to do. In last decade, researchers and scientific community have contributed lot for organization of individual’s inbox by classifying the emails into different categories. In this paper, we propose an intelligent mail box where email classification has been carried out on the basis of labels created by users and needs few training mails for future classification. Hence it provides more personalized mail box to the user. The proposed system has been tested with various classifiers, viz. Support Vector Machine, Naive Bayes, etc. and obtained the highest classification accuracy (66–100 %) with Naive Bayes.

Learning Based No Reference Algorithm for Dropped Frame Identification in Uncompressed Video

For last many years video authentication and detection of tampering in a video are major challenges in the domain of digital video forensics. This paper presents detection of one of the temporal tampering (frame drop) under no reference mode of tampering detection. Inspirit of the scheme presented by Upadhyay and Singh, this paper extends the features to train the SVM classifier and accordingly classify frames of given video as tampered or non-tampered frames, i.e. detects the tampering of frame drop. Subsequently given video is classified as tampered or non-tampered video. The obtained results with enhanced features show significant improvement in classification accuracy.

Data-parallel full reference algorithm for dropped frame identification in uncompressed video using genetic algorithm

In recent years due to easy availability of video editing tools, video sequences which are to be presented as evidence during court trials, can be tampered and therefore misguide the court proceeding. These video sequences are required to be authenticated by forensic experts before considering them as evidence during court trials. Frame drop is one of the most often temporal tampering and it is required to authenticate a video sequence against frame drop. Current work is an extension of authors pre-published work where they presented a full reference algorithm for dropped frame identification in uncompressed video using genetic algorithm which efficiently identifies dropped frame indices with accuracy ranging from 78 to 100 percentage under different test scenarios. To resolve the issue of scalability, this paper extends the algorithm by identifying set of independent tasks in pre-published FR algorithm and presented a data-parallel algorithm. We simulated the scheme over five video sequences (of duration 10 to 540 seconds) by dropping 0.5 percentage and 1 percentage video frames and analyzed the required processing time with 1, 2, 4, and 8 processors. Simulated results suggested that the presented algorithm is scalable and efficiently identifies dropped frame indices with average speedup of 1.77, 2.81, and 3.35 for 2, 4, and 8 processors respectively.

Dynamic simulator and planning tool for studying the defense and attack strategies in a war-zone

Augmented Reality (AR) helps create information-enriched, real-time environments that are close to the real world. This paper discusses a unique defense planning tool that has been developed as a technology demonstrator with systems and applications in AR. This tool has a direct link to the planning aspect of strategy development. It is expected that this will serve as a very useful tool for testing the concepts of defense strategic planning. It has features that can simulate an actual war-zone situation, which can help to test the strengths and weaknesses of the proposed defense strategies in real-time. Post simulation analysis of these strategies can help to modify the defense strategy (if needed) and strengthen the on-ground capacities and capabilities of the defense forces for mitigating the perceived threats from the enemy. It is a versatile tool, which can be further adapted to develop strategies that will help in minimizing the impact of natural calamities in zones which are prone to recurrent floods and seismic disturbances, thus contributing to improved disaster preparedness and management.

Spatio-Temporal extraction, visualization and investigation of worldwide disease alerts

This paper presents the architecture of a system for Spatio-temporal extraction, visualization and investigation (S.T.E.V.I.) of worldwide disease alerts based on geographical locations present in a text. Further, the framework and algorithms used for developing the system are described. Finally, the statistics of the application as well as the user-interface is discussed concluding with future scope at the end.