The object recognition concept is being widely used a result of increasing CCTV surveillance and the need for automatic object or activity detection from images or video. Increases in the use of various sensor networks have also raised the need of lightweight process frameworks. Much research has been carried out in this area, but the research scope is colossal as it deals with open-ended problems such as being able to achieve high accuracy in little time using lightweight process frameworks. Convolution Neural Networks and their variants are widely used in various computer vision activities, but most of the architectures of CNN are application-specific. There is always a need for generic architectures with better performance. This paper introduces the Dimension-Based Generic Convolution Block (DBGC), which can be used with any CNN to make the architecture generic and provide a dimension-wise selection of various height, width, and depth kernels. This single unit which uses the separable convolution concept provides multiple combinations using various dimension-based kernels. This single unit can be used for height-based, width-based, or depth-based dimensions; the same unit can even be used for height and width, width and depth, and depth and height dimensions. It can also be used for combinations involving all three dimensions of height, width, and depth. The main novelty of DBGC lies in the dimension selector block included in the proposed architecture. Proposed unoptimized kernel dimensions reduce FLOPs by around one third and also reduce the accuracy by around one half; semi-optimized kernel dimensions yield almost the same or higher accuracy with half the FLOPs of the original architecture, while optimized kernel dimensions provide 5 to 6% higher accuracy with around a 10 M reduction in FLOPs.
1 May 2020 in International Journal of Engineering Research & Technology
A Survey: Virtual, Augmented and Mixed Reality in Education - written by Sumit Patel , Binal Panchotiya , Akashkumar Patel published on 2020/06/01 download full article with reference data and citations
In today’s world, due to the rapid growth in music tracks via both offline and online and give better access to automatically classify its genre and based on that the recommendation for another song will impact a great experience to the user. So, this paper has proposed a system that uses deep learning approach for performing genre classification on the song and based on that the song will be recommended by using word2vec. For classification, it is important to obtain a large collection and index the songs accordingly to their genre and with the help of skip gram model, it will identify the similar context song for recommendation. So, the proposed system works as a whole music recommendation system for delivering great experience at user side.
The tumor also called as neoplasm, a growth in the abnormal tissue can be distinguished by its composition from the embracing tissue. A tumor prompts a cancer, a notable leading cause of death and responsible for about 13 % of all deaths worldwide. The incidence rate of cancer is rising at an unprecedented pace across the globe. Incredible radiological knowledge and expertise is required for accurate detection of tumors in medical imaging. Tumor detection automation is needed as there could be a shortage of professional radiologists during an era of unimaginable need. This paper examines the processes and techniques used as part of tumor detection in light of the results of magnetic resonance imaging (MRD in medical imaging. The techniques are considered based on computer vision can discern tumors in various kinds of medical imaging that aid with the diagnosis of myriad diseases, almost at a specialist level.
The generally accepted engineering design process involves generating requirements prior to developing concepts. Despite this, educators often observe that student engineers tend to have ideas for the eventual solution even before fully defining the problem. Thus, a method of exploiting this natural proclivity may result in a better overall product or process. Coevolutionary design, in which a problem and its solution are developed interdependently, may provide a theoretical construct for such a method. It may be that allowing for the creation of sketches as a first step may improve the resulting requirements, which in turn would enhance the result. To test this, an experiment was developed to observe the effects of rearranging the design sequence to use an early conceptual sketch in the elicitation of constraints and criteria. Requirements were generated by third- and fourth-year mechanical engineering students and were subsequently analyzed based on their variety, typology, and novelty. It was found that the use of a preliminary conceptual sketch had a significant (positive) effect on the typology and novelty of the resulting requirements (α = 0.05), though no change in their variety was observed. Also, an additional intermediate step of identifying key features in the sketch further influenced requirement characteristics. The findings of this study support the coevolutionary model of design and suggest that the sketching of ideas and the identification of features in advance of listing requirements may be a valid design practice in the future. By creating an initial concept sketch and then distinguishing the important aspects of that sketch, engineers would be able to extract applicable requirements which they find inherent in their initial ideas. Finally, this method can align more naturally with the approach that many students employ in design. Rather than modifying behaviors, this method can exploit student behaviors to positive effect.
Autonomous exploration and mapping of unknown subterranean areas using fleet of Micro Aerial Vehicles (MAVs) have achieved great interest in the past decade. The application of such research direction is extended beyond mining industry for example the exploration of potential lava tubes and caves on Mars using Martian MAVs. In this article we present a Collaborative Exploration and Mapping approach for exploration of large scale multi-branched virtual cave environment using multiple MAVs. An important aspect of such approach is to establish a decent collaboration with minimal information exchange among agents such that the exploration area can be maximized by preventing multiple agents covering overlapping parts of the environment. The collaborative exploration approach in this article is based on sharing local traversable frontier points as information among agents such that the each agent can identify the target areas in order to extend global map in an efficient manner. This work proposes a novel baseline collaborative exploration approach for optimal target area selection with a resource constrained MAVs in order to efficiently as well as rapidly explore subterranean environment. Extensive simulations are performed to validate the proposed collaborative strategy in order to establish a multi agent coordinated exploration-coverage mission.