Manuscripts

Recent Papers

Research Paper

Natural scene image classification using CNN

Research mainly focused on CNN model for feature extraction and classification of Images. Convolutional Neural Network (CNN) has demonstrated promising performance in image classification tasks. In this project, the algorithm is used to classify the images or natural scenes into 6 classes. This model at last predicts the accuracy or probabilities of different class labels and this probability is used for the predicting class at the end. This dataset is used for both training and testing purpose. It provides the accuracy rate 84.93%. Images with combination of two scenes creates and ambiguity hence it is difficult for model to classify. Therefore, it leads to failure in algorithm sometimes. Images used in the training purpose are RGB images. The computational time for processing these images is relatively high as compare to other normal images. Stacking the model with more layers and training the network with more image data using clusters of GPUs provide more accurate results of classification of images.

Published by: Jayanth H. N.

Author: Jayanth H. N.

Paper ID: V6I6-1184

Paper Status: published

Published: December 4, 2020

Full Details
Research Paper

Identification of strategic networks in entrepreneurial networking process: A case study of influencer entrepreneur in fashion startups, Thailand

From social media influencers (SMIs) who gain social capital in the form of follower fan base on social media platforms, developing further career opportunities into a successful entrepreneur. This significant phenomenon has seen the emergence of influencer entrepreneurs utilizing social media platforms to disclose their own personal products or services. Underlying this growing trend is the dynamic interplay of networks and the firm activities. In order to better understand in specify research area, the importance of network dimensions have been conceptualized in start up process. To explore entrepreneurial networking process in which driving influencer entrepreneur into entrepreneurial success. The key implication of the research is the strategic network for achieving entrepreneurial success in firm performance and superior network outcomes. A qualitative research approach enabled triangulated data investigation with both primary and secondary sources to facilitated the emergence of relevant theme; Procurement and supplier relations, research design and development of product, human resource planning and training, management technology and management system , warehouse and logistics distribution, marketing intelligence and marketing (Fashion retailing, marketing and merchandising) and after-sale service and cash collections. The key strategies in fashion start up networking process influencer entrepreneur that uncovered in the study can thereby leveraging chances to turn start ups into success and sustainability and enhanced some network dimensions on how to become more competitive and successful in the future.

Published by: Nipaporn Promthong, Chia-Han,Yang

Author: Nipaporn Promthong

Paper ID: V6I6-1177

Paper Status: published

Published: November 28, 2020

Full Details
Research Paper

Optimal load forecasting by hybrid the artificial neural network and firefly algorithm

In the smart grid, load forecasting algorithms used for estimating the electricity demand based on historical data. It helps in generating accurate electricity and overcoming the two challenges such as a shortage of electricity and excess generation cost. In the literature, various traditional load forecasting algorithms proposed to predict the electricity demand but never the accurate results. Therefore, advanced algorithms come into the picture such as artificial intelligence algorithms. In this paper, we have hybrid the Artificial Neural Network (ANN) and firefly algorithm for load prediction. Initially, the ANN algorithm is trained based on the historical data then applied to it. After that, the firefly algorithm is used for searching for the optimal learning rate for ANN. The experimental results are performed in the MATLAB 2015a. We have measured various performance analysis parameters and compared with the existing results. From the study, we found that the proposed algorithm gives better accuracy as compared to the existing algorithms

Published by: Rishav, Puneet Jain, Chakshu Goel

Author: Rishav

Paper ID: V6I6-1170

Paper Status: published

Published: November 27, 2020

Full Details
Review Paper

Artificial intelligence and robotics: The enhanced paediatric dentist

Automation would be an inevitable course of manual work facing every field in the future in view of technological advancements. The mental and physical strain incumbent on the paediatric dentist due to long hours of managing the child and the unwavering focus that the trade demands may compromise the quality of service. A system capable of physical manipulation that is powered by an intelligent program would be the ideal assistant to the dentist for carrying out technique sensitive procedures. Data management, diagnosis, treatment planning and student education can transcend to a new plane of execution with Artificial Intelligence and Robot enhanced Paediatric Dentist at the epicentre. Among the many hurdles faced by the idea of turning robotics in dentistry into a tangible reality is the extreme cost and bureaucratic resistance.

Published by: Chandra Kanth B., Chandradeep, Swapna Manepalli

Author: Chandra Kanth B.

Paper ID: V6I6-1150

Paper Status: published

Published: November 27, 2020

Full Details
Research Paper

Smart image enhancement technique by removal of undesirable objects/background from the image

The devil is always in your picture. You always miss the best shot and that particular tiny object in your photo can worsen the entire composition and result in something opposite from what you hoped for. The proposed AI based image enhancement model will first detect unwanted objects/background based on the training of large set of already edited images by users then intelligently reconstructs the image without those objects. We have prepared a dataset with distracting elements in the images and used it to train our predictor model which predicts the distracting regions and thereafter used image-inpainting to remove those areas which results in a standalone system for distractor removal with no user input. In the proposed method of this paper, the image is first segmented using Convolutional networks for semantic segmentations and then each segment is classified in terms of the score of distractors on the basis of various features which almost covers all types of distractors in an image. Our main focus was to collect the data which contains all kinds of distractors and then deciding the features which classify an object as a distractor in an image. Detection and removal of distracting regions helps to enhance the beauty and visual quality of the image which can be fulfilled by using our model.

Published by: Pankaj Tanwar, Karishma Kumari, Saqib Kamal

Author: Pankaj Tanwar

Paper ID: V6I6-1161

Paper Status: published

Published: November 27, 2020

Full Details
Research Paper

Speech Recognition for Crime Analysis

Human-Computer Interaction (HCI) is a multidisciplinary field of study focusing on the design of computer technology and, in particular, the interaction between humans (the users) and computers. While initially concerned with computers, HCI has since expanded to cover almost all forms of information technology design. We aim to take this further in the field of crime analysis by developing a system for speech recognition. This will be a huge contribution to the crime department where speech recognition will not only help them to handle large amount of data but also in checking the criminal records faster. This is a modern and advanced technique where with the help of voice recognition test file will be compared with database files.

Published by: Mehak Sharma

Author: Mehak Sharma

Paper ID: V6I6-1153

Paper Status: published

Published: November 16, 2020

Full Details