(SoE) School of Engineering

Shiv Nadar University is a comprehensive, multidisciplinary, research-focused, and student-centric University offering a full range of academic programs at the undergraduate, postgraduate and doctoral level. With state-of-the-art infrastructure, the University comprises of academic wings, sophisticated labs, international standard sports facilities, amphitheaters, auditorium, conference rooms and smart classrooms. The University&rsquo;s goal is to become internationally recognized for the quality of its research and creative endeavors and their applicability to improving quality of life, generating new insights and expanding the boundaries of human knowledge creativity. Committed to excellence in teaching, research and service, the University aims to serve the higher education needs of India and the world beyond.

Shiv Nadar University

Recent years have been witnessing significant developments in research in human action recognition based on skeleton data. The graphical representation of the human skeleton, available with the dataset, provides opportunity to apply Graph Convolutional Networks (GCN), to avail efficient analysis of deep spatial-temporal information from the joint and skeleton structure. Most of the current works in skeleton action recognition use the temporal aspect of the video in short-term sequences, ignoring the long-term information present in the evolving skeleton sequence. The proposed long-term Spatio-temporal Contrastive Learning framework for Skeleton Action Recognition uses an encoder-decoder module. The encoder collects deep global-level (long-term) information from the complete action sequence using efficient self-supervision. The proposed encoder combines knowledge from the temporal domain with high-level information of the relative joint and structure movements of the skeleton. The decoder serves two purposes: predicting the human activity and predicting skeleton structure in the future frames. The decoder primarily uses the high-level encodings from the encoder to anticipate the action. For predicting skeleton structure, we extract an even deeper correlation in the Spatio-temporal domain and merge it with the original frame of the video. We apply a contrastive framework in the frame prediction part so that similar actions have similar predicted skeleton structure. The use of the contrastive framework throughout the proposed model helps achieve exemplary performance while employing a self-supervised aspect to the model. We test our model on the NTU-RGB-D-60 dataset and achieve state-of-the-art performance. The codes related to this work are available at: \url{https://github.com/AnshulRustogi/Long-Term-Spatio-Temporal-Framework}.

World University root node

Computer Science and Engineering

IJCNN

Long-term Spatio-temporal Contrastive Learning framework for Skeleton Action Recognition

The efficacy of Action Recognition methods depends upon relevance of the action (Verb) with respect to the subject (Noun). Existing methods overlook the suitability of Noun-Verb combination in defining an action. In this work, we propose an algorithm called Reduced Verb Set Generator (RVSGen) to reduce the number of possible verbs related to the actions, based upon the relevance of noun-verb combination. A dual modal fusion model for egocentric activity recognition is proposed here to combine the features extracted from the RGB channels and the Optical flow vectors for an egocentric video to recognize the human activity. Unlike state-of-the-art methods, where the key objects and temporal cues are extracted simultaneously, the proposed model first extracts the spatial features i.e., object information (Noun) from the RGB channels and then relates the Noun with the suitable action (Verb) obtained from motion information (Optical Flow). The verbs are predicted by an ConvLSTM architecture with the help of a modified softmax. The notion behind the modified softmax is to estimate the probability distribution with the reduced verb set obtained from RVSGen and the feature vector obtained from the ConvLSTM. With the help of an end-to-end trained architecture, the noun and verb are predicted which are then concatenated together constituting an action. The experiments are performed on a benchmark dataset. The results show the efficacy of the proposed method compared to the state-of-the-art. The codes related to this work can be found at: \url{https://github.com/mpLogics/EgoAR-RVSGen}.

First-person Activity Recognition by Modelling Subject-Action Relevance

Single image dehazing is a popular research topic among the researchers in computer vision, machine learning, image processing, and graphics. Most of the recent methods for single image dehazing are based upon supervised learning set up. However, supervised methods require annotation of the data, which often makes the dehazing methods biased towards the manual annotation errors. Unsupervised methods are more likely to produce realistic, clear images. However, fewer efforts are found in the literature for single image dehazing in unsupervised set up. We propose an enhanced CycleGAN architecture for Unpaired single image dehazing, with an attention-based transformer architecture embedded in the generator. The proposed transformer comprises three components: 1) A Feature Attention (FA) block combining channel attention and pixel attention mechanism, 2) A Dynamic feature enhancement block for dynamically capturing the spatial structured features and 3) An adaptive mix-up module to preserve the flow of shallow features from downsampling. Experiments on the benchmark datasets show the efficacy of the proposed method. Codes for this work are available in the link: \url{https://github.com/rsjai47/Attention-Based-CycleDehaze}.

Attention-based Single Image Dehazing Using Improved CycleGAN

We propose a novel video sampling scheme for human action recognition in videos, using Gaussian Weighing Function. Traditionally in deep learning-based human activity recognition approaches, either a few random frames or every $k^{th}$ frame of the video is considered for training the 3D CNN, where $k$ is a small positive integer, like $4$, $5$, or $6$. This kind of sampling reduces the volume of the input data, which speeds-up the training network and also avoids overfitting to some extent, thus enhancing the performance of the 3D CNN model. In the proposed video sampling technique, consecutive $k$ frames of a video are aggregated into a single frame by computing a Gaussian-weighted summation of the $k$ frames. The resulting frame preserves the information in a better way than the conventional approaches and experimentally shown to perform better. In this paper, a 3-Dimensional deep CNN is proposed to extract the spatio-temporal features and follows Long Short-Term Memory (LSTM) to recognize human actions. The proposed 3D CNN architecture is capable of handling the videos where the camera is placed at a distance from the performer. Experiments are performed with KTH, WEIZMANN, and CASIA-B Human Activity and Gait datasets, whereby it is shown to outperform state-of-the-art deep learning based techniques. We achieve 95.78%, 95.27%, and 95.27% over the KTH, WEIZMANN, and CASIA-B human action and gait recognition datasets, respectively. &nbsp;&nbsp;

Multimedia Tools and Applications

An Information-rich Sampling Technique over Spatio-Temporal CNN for Classification of Human Actions in Videos

Micro-expression has emerged as a promising modality in affective computing due to its high objectivity in emotion detection. Despite the higher recognition accuracy provided by the deep learning models, there are still significant scope for improvements in micro-expression recognition techniques. The presence of micro-expressions in small-local regions of the face, as well as the limited size of available databases, continue to limit the accuracy in recognizing micro-expressions. In this work, we propose a facial micro-expression recognition model using 3D residual attention network named MERANet to tackle such challenges. The proposed model takes advantage of spatial-temporal attention and channel attention together, to learn deeper fine-grained subtle features for classification of emotions. Further, the proposed model encompasses both spatial and temporal information simultaneously using the 3D kernels and residual connections. Moreover, the channel features and spatio-temporal features are re-calibrated using the channel and spatio-temporal attentions, respectively in each residual module. Our attention mechanism enables the model to learn to focus on different facial areas of interest. The experiments are conducted on benchmark facial micro-expression datasets. A superior performance is observed as compared to the state-of-the-art for facial micro-expression recognition on benchmark data.

ICVGIP

MERANet: Facial Micro-Expression Recognition using 3D Residual Attention Network

The treatment of domestic sewage using a moving bed hybrid <a href="https://www.sciencedirect.com/topics/earth-and-planetary-sciences/bioreactor">bioreactor</a> (MBHBR) was investigated by varying the concentrations of <a href="https://www.sciencedirect.com/topics/earth-and-planetary-sciences/chemical-oxygen-demand">chemical oxygen demand</a> (COD) (156&nbsp;±&nbsp;14-1141&nbsp;±&nbsp;78 mg/L), ammonia (NH3-N) (2&nbsp;±&nbsp;0.08-48.24&nbsp;±&nbsp;0.93 mg/L), phosphorus (PO4-P) (97&nbsp;±&nbsp;3.2-281&nbsp;±&nbsp;2.65 mg/L), and hydraulic retention times (HRTs). Initially, the reactor was operated at start-up mode (COD: 156&nbsp;±&nbsp;14 mg/L), and gradually the COD concentrations were increased from phase-I (257&nbsp;±&nbsp;24 mg/L) to phase-V (1141&nbsp;±&nbsp;78 mg/L) by reducing the HRTs from 48 to 3 h. A significant improvement in COD removal, maximum up to 96%, was observed at a concentration of 858&nbsp;±&nbsp;51 mg/L (phase-IV), then a sudden drop up to 81% was observed at a higher concentration of 1141&nbsp;±&nbsp;78 mg/L (phase-V). Due to the increase in COD concentrations, average NH3-N removal was dropped from 85 to 32% from phase-I to Phase-V, respectively. In addition, the stability of MBHBR was studied to understand its performance in shut-down and shock loading phases. The co-treatment of sewage and leachate was also performed in MBHBR (COD: 864-881 mg/L and HRT: 6 h), which showed a COD removal of 77%–80% and NH3-N removal of 60%–94%. Proteobacteria was identified as a predominant species in MBHBR, which played a significant role in the co-treatment of sewage and leachate. The maximum biodegradation rate (<math xmlns="http://www.w3.org/1998/Math/MathML"><msub is="true"><mrow is="true"><mi is="true">r</mi></mrow><mrow is="true"><mi mathvariant="normal" is="true">max</mi></mrow></msub></math>) and half-saturation constant (<math xmlns="http://www.w3.org/1998/Math/MathML"><msub is="true"><mrow is="true"><mi is="true">K</mi></mrow><mrow is="true"><mi mathvariant="normal" is="true">s</mi></mrow></msub></math>) were estimated as 13.3 mg/L h and 361.8 mg/L, respectively.

Civil Engineering

Fulltext

Environmental Technology & Innovation

<a href="https://www.sciencedirect.com/science/article/pii/S2352186421006465">Treatment of domestic sewage and leachate using a moving bed hybrid bioreactor</a>

Phubbing is defined as ignoring people with whom you have a face-to-face relationship to attend to smartphones. The phenomenon of phubbing particularly affects the teenage and young segments of the population. The main problem lies in the impact it has on individuals’ social relationship. A lack of validated instruments to diagnose this phenomenon has been observed amongst the Spanish-speaking youth. The objective pursued with this research was to analyse the structural validity and reliability of the Spanish scale in a sample of 454 Peruvian university students. A reliability study was carried out following Cronbach and McDonald, complemented with an exploratory and confirmatory factor analysis. The results show good reliability and validity values. Finally, some aspects associated with users’ profiles in relation to the scale were discussed too. A need exists to have adapted instruments which permit to measure emerging social threats such as phubbing, so that risk profiles can be identified and for us to be able to act in time. Most of the students surveyed regularly engaged in phubbing, and a significant percentage of them had personal and social problems because of this, including lack of sleep hours or arguments with friends and relatives, to quote but two. &nbsp;

Electrical Engineering

NAER

<a href="https://naerjournal.ua.es/article/view/v10n2-1">Scale for Measuring Phubbing in Peruvian University Students: Adaptation, Validation and Results of Its Application</a>

The crop diseases can’t detected accurately by only analysing separate disease basis. Only with the help of making comprehensive analysis framework, users can get the predictions of most expected diseases. In this research, IOT and machine learning based technique capable of processing acquisition, analysis and detection of crop health information in the same platform is introduced. The proposed system supports distinguished services by monitoring crop and also managed its data, devices and models. This system also supports data sharing and communication with the help of IOT using unmanned aerial vehicle (UAV) and maintains high communication standards even in bad communication environment. Therefore, IOT and machine learning ensures the high accuracy of disease prediction in crop. The proposed integrated system is capable of detecting health of crop through analysis of multi-spectral images captured through the IOT associated UAV. The various machine learning is also applied to test the performance of our system and compared with the existing disease detection methods.&nbsp;

Journal of Scientific and Industrial Research

Detecting Crop Health using Machine Learning Techniques in Smart Agriculture System

Early action prediction in video is a challenging task where the action of a human performer is expected to be predicted using only the initial few frames. We propose a novel technique for action prediction based on Deep Reinforcement learning, employing a Deep Q-Network (DQN) and the ResNext as the basic CNN architecture. The proposed DQN can predict the actions in videos from features extracted from the first few frames of the video, and the basic CNN model is adjusted by tuning the hyperparameters of the CNN network. The ResNext model is adjusted based on the reward provided by the DQN, and the hyperparameters are updated to predict actions. The agent's stopping criteria is higher or equal to the validation accuracy value. The DQN is rewarded based on the sequential input frames and the transition of action states (i.e., prediction of action class for an incremental 10 percent of the video). The visual features extracted from the first 10 percent of the video is forwarded to the next 10 percent of the video for each action state. The proposed method is tested on the UCF101 dataset and has outperformed the state-of-the-art in action prediction.

Early Prediction of Human Action by Deep Reinforcement Learning

Social Media these days has become the prime medium for religious debates. While some people utilise the medium wisely to talk about the rational thoughts associated with different religions, others might use it to marginalize or dehumanize certain religions. Therefore, there is a dire need to analyse content having a religious context so that appropriate actions can be taken by the social media or blogging platforms against people attempting to propagate unhealthy religious messages. Therefore, in this paper, a multi-module machine learning based algorithm is proposed to analyse the religious context in online social media content. To do that, data has been gathered from popular social media sites Twitter. A methodology is then developed to identify tweets that talk about religion. Once, it is detected if the tweet is religion themed or not, sentiment analysis is performed and then the nature of the tweets (offensive or not) is detected to get a deeper understanding of the content that is being posted online. Based on the evaluation of various classification models, it is found that Naive Bayes performs the best at detecting a religious theme in tweets, achieving an accuracy of 84%. On the other hand, XGBoost performs the best at detection of offensive content in the domain of religion. Our results identify online religious offensive content and offer a broader understanding of the phenomenon, providing directions for prevention and detection approaches.&nbsp;