Conference Papers
Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/123456789/28506
Browse
15 results
Search Results
Item Selfie Detection by Synergy-Constraint Based Convolutional Neural Network(Institute of Electrical and Electronics Engineers Inc., 2017) Annadani, Y.; Naganoor, V.; Jagadish, A.K.; Chemmangat, K.Categorisation of huge amount of data on the multimedia platform is a crucial task. In this work, we propose a novel approach to address the subtle problem of selfie detection for image database segregation on the web, given rapid rise in the number of selfies being clicked. A Convolutional Neural Network (CNN) is modeled to learn a synergy feature in the common subspace of head and shoulder orientation, derived from Local Binary Pattern (LBP) and Histogram of Oriented Gradients (HOG) features respectively. This synergy was captured by projecting the aforementioned features using Canonical Correlation Analysis (CCA). We show that the resulting network's convolutional activations in the neighbourhood of spatial keypoints captured by SIFT are discriminative for selfie-detection. In general, proposed approach aids in capturing intricacies present in the image data and has the potential for usage in other subtle image analysis scenarios apart from just selfie detection. We investigate and analyse the performance of the popular CNN architectures (GoogleNet, Alexnet), used for other image classification tasks, when subjected to the task of detecting the selfies on the multimedia platform. The results of the proposed approach are compared with these popular architectures on a dataset of ninety thousand images comprising of roughly equal number of selfies and non-selfies. Experimental results on this dataset shows the effectiveness of the proposed approach. © 2016 IEEE.Item Unobtrusive students' engagement analysis in computer science laboratory using deep learning techniques(Institute of Electrical and Electronics Engineers Inc., 2018) Ashwin, T.S.; Guddeti, R.M.Nowadays, analysing the students' engagement using non-verbal cues is very popular and effective. There are several web camera based applications for predicting the students' engagement in an e-learning environment. But there are very limited works on analyzing the students' engagement using the video surveillance cameras in a teaching laboratory. In this paper, we propose a Convolutional Neural Networks based methodology for analysing the students' engagement using video surveillance cameras in a teaching laboratory. The proposed system is tested on five different courses of computer science and information technology with 243 students of NITK Surathkal, Mangalore, India. The experimental results demonstrate that there is a positive correlation between the students' engagement and learning, thus the proposed system outperforms the existing systems. © 2018 IEEE.Item A Transfer Learning Approach for Diabetic Retinopathy Classification Using Deep Convolutional Neural Networks(Institute of Electrical and Electronics Engineers Inc., 2018) Krishnan, A.S.; Clive, D.R.; Bhat, V.; Ramteke, P.B.; Koolagudi, S.G.Diabetic Retinopathy is a disease in which the retina is damaged due to diabetes mellitus. It is a leading cause for blindness today. Detection and quantification of such mellitus from retinal images is tedious and requires expertise. In this paper, an automatic identification of severity of Diabetic Retinopathy using Convolutional Neural Networks (CNNs) with a transfer learning approach has been proposed to aid the diagnostic process. A comparison of different CNN architectures such as ResNet, Inception-ResNet-v2 etc. is done using the quadratic weighted kappa metric. The qualitative and quantitative evaluation of the proposed approach is carried out on the Diabetic Retinopathy detection dataset from Kaggle. From the results, we observe that the proposed model achieves a kappa score of 0.76. © 2018 IEEE.Item Power Quality Event Classification Using Transfer Learning on Images(Institute of Electrical and Electronics Engineers Inc., 2019) Manikonda, S.K.G.; Santhosh, J.; Sreckala, S.P.K.; Gangwani, S.; Gaonkar, D.N.Given the ever-increasing complexity of the electrical grid system, power quality events have been surging in frequency with each passing day. Due to their potential to cause massive losses for a wide variety of customers, it is crucial that such events are detected and classified immediately for appropriate response. in this paper, a novel approach has been developed wherein Transfer Learning techniques have been employed to classify power quality events using image classification. More specifically, the VGG16 model has been utilized to classify five distinct power quality issues by using scalograms as input images. 489 scalograms were generated via feature extraction using wavelet transforms. The VGG16 model has then been trained and tested using the same. Thereafter, the model performance has been evaluated, and the results have been discussed. © 2019 IEEE.Item Identifying Parking Lots from Satellite Images using Transfer Learning(Institute of Electrical and Electronics Engineers Inc., 2019) Kumar, S.; Thomas, E.; Horo, A.With the advent of digital image processing techniques and convolutional neural networks, the world has derived numerous benefits such as computerized photography, biological Image Processing, finger print and iris recognition, to name a few. Computer vision coupled with convolutional neural networks has attributed machines with a virtual intellectual ability to recognize and distinguish images based on several characteristics that may be impossible for the human eye to perceive. We have exploited this advancement in technology to particular use case of detecting number of empty and occupied parking slots from satellite images of parking lots. We have proposed a befitting sequence of classical image processing techniques and algorithms to perform pre-processing of satellite images of parking spaces. Moreover, we have proposed a Convolutional Neural Network model that takes as input these preprocessed images and identifies the empty and occupied parking slots with an accuracy of 97.73%. The potential benefits of using Neural Networks to realize the objective can be extended to open parking spaces of different configurations. This is due to the fact that establishing sensors over a large number of parking slots over a given open parking space can be a cumbersome and exorbitant task. The proposed model comprises of few convolutional layers and uses Rectified Linear Classification activation function. © 2019 IEEE.Item Empirical Study on Multi Convolutional Layer-based Convolutional Neural Network Classifier for Plant Leaf Disease Detection(Institute of Electrical and Electronics Engineers Inc., 2020) Sunil, C.K.; Jaidhar, C.D.; Patil, N.Recognizing the plant disease automatically in real-time by examining a plant leaf image is highly essential for farmers. This work focuses on an empirical study on Multi Convolutional Layer-based Convolutional Neural Network (MCLCNN) classifier to measure the detection efficacy of MCLCNN on recognizing plant leaf image as being healthy or diseased. To achieve this, a set of experiments were conducted with three distinct plant leaf datasets. Each of the experiments were conducted by setting kernel size of 3× 3 and each experiment was conducted independently with different epochs i.e., 50, 75, 100, 125, and 150. The MCLCNN classifier achieved minimum accuracy of 87.47% with 50 epochs and maximum accuracy of 99.25% with 150 epochs for the Peach plant leaves. © 2020 IEEE.Item Comparing CNNs and GANs for Image Completion(Institute of Electrical and Electronics Engineers Inc., 2021) Saji, R.; Anand, S.K.; Chandavarkar, B.R.Imperfections or defects inevitably occur in images due to inexperienced photographers, inadequate methods of preservation, or even some deliberate hacking. Image restoration or completion has been performed using various manual methods in the past, be it being drawn by artists based on their creativity or deleting noise and blur effects using software like Photo-shop. On a large scale, manual image completion is infeasible and has quite a lot of limitations. Modern advancements in Computer Vision and Deep Learning have allowed man to automate such tasks with high efficiency. Manual restoration usually relies on prior experience in the subject and sometimes even creativity to reconstruct the image based on the artist's imagination. At the same time, deep learning produces excellent results given enough training data. Deep learning methods can improvise and generalize better too and hence outperform the traditional manual methods. In this project, image completion is performed using 2 Deep Learning models - Convolutional Neural Networks(CNN) and Generative Adversarial Networks(GAN). Adversarial Networks have been proven to be very handy in image to image translation tasks and image reconstruction and hence this is explored widely. Both Deep Convolutional GANs as well as Conditional GANs are used for this task and their respective performances are compared for the above task. © 2021 IEEE.Item Effect of Different Color Spaces on Deep Image Segmentation(Institute of Electrical and Electronics Engineers Inc., 2021) Sushma, B.; Pulikala, P.Image segmentation is an important application in computer vision, proposed to partition an image into meaningful regions on a specific criterion. In recent days, image segmentation tasks have achieved state of the art performance using deep neural and fully connected networks. The datasets used for the segmentation task mainly consist of image data in RGB color space and the deep segmentation architectures are trained without modifying the color space. In this study, the importance of color space is investigated and the obtained results show that the color space can affect the segmentation performance remarkably. Certain regions of interest in images belonging to a particular domain can be segmented better when represented in a certain form of color space. To explore on this two datasets from medical and satellite imagery are considered. The UNET model is modified to accept images as a combination of color spaces and is trained to segment the colonoscopy images for polyps and satellite images for roads under individual and combination of color spaces. Experiments show that the performance of polyp segmentation is better when a combination of HSV+YCbCr color space is considered. Road segmentation in satellite imagery is better in LAB+HSV color space. © 2021 IEEE.Item Automated Summarization of Gastrointestinal Endoscopy Video(Springer Science and Business Media Deutschland GmbH, 2023) Sushma, B.; Aparna., P.Gastrointestinal (GI) endoscopy enables many minimally invasive procedures for diagnosing diseases such as esophagitis, ulcer, polyps and cancers. Guided by the endoscope’s video sequence, a physician can diagnose the diseases and administer the treatment. Unfortunately, due to the huge amount of data generated, physicians are currently discarding procedural video and rely on a small number of carefully chosen images to record a procedure. In addition, when a patient seeks a second opinion, the assessment of lesions in a huge video stream necessitates a thorough examination, which is a time-consuming process that demands much attention. To reduce the length of the video stream, an automated method to generate the summary of endoscopy video recordings consisting only of abnormal frames by using deep convolutional neural networks trained to classify normal, abnormal and uninformative frames is proposed. Results show that our method can efficiently detect abnormal frames and is robust to the variations in the frames. The proposed CNN architecture outperforms the other classification models with an accuracy of 0.9698 with less number of parameters. © IFIP International Federation for Information Processing 2023.Item A Paradigm Shift in Brain Tumor Classification: Harnessing the Potential of Capsule Networks(Institute of Electrical and Electronics Engineers Inc., 2023) Raythatha, Y.; Vani, M.Accurate and timely classification of brain tumors is critical for developing effective treatment plans and predicting treatment outcomes. However, CNN-based models commonly used for this task have limitations, such as their reliance on large amounts of training data and difficulties with input orientation and transformations. To address these limitations, we propose a CapsNet-based model for brain tumor classification designed to effectively handle limited datasets, class imbalance, and input transformations. CapsNet relies on 'capsules,' groups of neurons that work together to represent specific input image features and are resistant to input orientation and transformations. Our study compares the performance of the proposed CapsNet-based model with state-of-The-Art CNN models, and our results demonstrate that the CapsNet-based model outperforms CNN models in terms of accuracy and robustness to input orientation and transformations. These findings suggest that CapsNet has the potential to be a promising alternative to CNNs for accurate and efficient brain tumor classification. © 2023 IEEE.