MIT-World Peace University (MIT-WPU) is a top educational institute located in Pune, India. It is recognized within&nbsp;UGC under Government of Maharashtra Act No. XXXV 2017.&nbsp;

It is part of the MIT group of institutions (established in 1983) which covers 10+ campuses across India and has more than 50,000+ students enrolled across various disciplines at a given point of time. The institute houses 15 schools ranging from Engineering &amp; Technology, Design, Pharmacy, Commerce, Governance among others. It offers undergraduate, postgraduate and PhD programs among different disciplines.&nbsp;

&nbsp;

MIT World Peace University

Humans can perform an enormous number of actions like running, walking, pushing, and punching, and can perform them in multiple ways. Hence recognizing a human action from a video is a challenging task. In a supervised learning environment, actions are first represented using robust features and then a classifier is trained for classification. The selection of a classifier does affect the performance of human action recognition. This work focuses on the comparison of two structures of the neural network, namely, feed forward neural network and cascade forward neural network, for human action recognition. Histogram of oriented gradients (HOG) and histogram of optical flow (HOF) are used as features for representing the actions. HOG represents the spatial features of the video while HOF gives motion features of the video. The performance of two neural network architectures is compared based on recognition accuracy. Well-known publically available datasets for action and interaction detection are used for testing. It is seen that, for human action recognition applications, feed forward neural network gives better results in terms of higher recognition accuracy than Cascade forward neural network. © 2022 Institute of Advanced Engineering and Science. All rights reserved.

School of Computer Engineering & Technology

Engineering and Technology

Indonesian Journal of Electrical Engineering and Computer Science

Comparison of feed forward and cascade forward neural networks for human action recognition

Local feature-based approaches are proven to be most successful for the application of human action recognition. This work aims to study the effect of the number of cuboids chosen on the performance of human action recognition. A 3D region, called cuboid, is extracted around every spatio-temporal interest point and an entropy-based cuboid selection method is implemented for choosing the cuboids having maximum information. The proposed method is evaluated using UT interaction and SBU kinect interaction datasets. Results show that maximum classification accuracy and F1 Score are obtained when the top 40 cuboids are selected for the feature extraction. The classification accuracy and F1 Score increase when the number of selected cuboids is increased from 10 to 40 and remain constant when the number is increased beyond 40. The selected 40 cuboids are then used for feature extraction. To carry out this experiment, wavelet coefficients extracted from selected cuboids are used as features. Copyright © 2022 Inderscience Enterprises Ltd.

International Journal of Intelligent Systems Technologies and Applications

Performance evaluation of entropy-based cuboid selection technique for human interaction recognition

In the current era where ample data is generated every second and available to every individual in the form of information, it is easy to gain knowledge from this information but a systematic and methodical approach is required when it comes to skill development. Online and on campus programs help in gaining knowledge but the skills to apply this knowledge needs to be developed gradually through appropriate teaching pedagogy which helps in practicing higher order thinking skills as proposed by Bloom and well known as Bloom’s Taxonomy. In conventional teaching learning approach adopted for online and in-person classroom sessions, the transmission of knowledge takes place most of the time where learner’s involvement may be quite less. So from the perspective of improvement in learner’s involvement in teaching learning process, active learning methods play a significant role and hence practiced widely in recent times. Variety of methods are recommended by scientists and Education experts to improve upon assimilation of knowledge for problem solving. Depending upon the nature of the problem the learner either needs to use one of the lower order thinking skills like apply or higher order thinking skills like analyse, evaluate and create as defined by Bloom. The study presented in this paper discusses case studies of practicing active learning techniques and developing problem solving skills through these techniques. The consolidated feedback analysis for the active learning techniques practiced show that the 96% learners have agreed for clarity of understanding, 92.72% learners have shown confidence in application of concepts and 97.27% participants have shown improved satisfaction in learning. Qualitative testimonies suggests the promising strength of PBL and effectiveness of it in improvement of problem solving skills. The study encourages the effective use of active learning techniques for skill development which is need of time. © 2021, Rajarambapu Institute Of Technology. All rights reserved.

Journal of Engineering Education Transformations

Role of Active Learning Techniques in Development of Problem Solving Skills

The integration of the machine learning techniques in healthcare can be of huge benefit aimed at curing illness of millions of people. A lot of effort has been taken by researchers to detect and provide early-stage insights into cancer diagnosis. In the machine learning research community, various algorithms-KNN, SVM, Decision Trees and Random Forest have been applied to calculate the presence or decisiveness of cancer in correspondence with the symptoms shown by the patients. This paper aims to analyze the symptoms of the different age groups Youth, Working Class and Elderly. Tree-based algorithms like Decision Trees, Random Forest and XGBoost have been used to identify the underlying data patterns in order to calculate relative feature importances. It has been concluded that Coughing of Blood, Clubbing of Finger Nails, Genetic Risk, Passive Smoking and Snoring are the factors that are responsible for lung cancer in all the age groups in most of the cases. © 2020 IEEE.

Proceedings of the 3rd International Conference on Intelligent Sustainable Systems, ICISS 2020

Symptom Analysis using a Machine Learning approach for Early Stage Lung Cancer

In this paper, a new local feature, called, Salient Wavelet Feature with Histogram of Oriented Gradients (SWFHOG) is introduced for human action recognition and behaviour analysis. In the proposed approach, regions having maximum information are selected based on their entropies. The SWF feature descriptor is formed by using the wavelet sub-bands obtained by applying wavelet decomposition to selected regions. To improve the accuracy further, the SWF feature vector is combined with the Histogram of Oriented Gradient global feature descriptor to form the SWFHOG feature descriptor. The proposed algorithm is evaluated using publicly available KTH, Weizmann, UT Interaction, and UCF Sports datasets for action recognition. The highest accuracy of 98.33% is achieved for the UT interaction dataset. The proposed SWFHOG feature descriptor is tested for behaviour analysis to identify the actions as normal or abnormal. The actions from SBU Kinect and UT Interaction dataset are divided into two sets as Normal Behaviour and Abnormal Behaviour. For the application of behaviour analysis, 95% recognition accuracy is achieved for the SBU Kinect dataset and 97% accuracy is obtained for the UT Interaction dataset. Robustness of the proposed SWFHOG algorithm is tested against Camera view angle change and imperfect actions using Weizmann robustness testing datasets. The proposed SWFHOG method shows promising results as compared to earlier methods.

World University root node

Fulltext

International Journal of Advanced Computer Science and Applications

A Novel Human Action Recognition and Behaviour Analysis Technique using SWFHOG

Extracting keyframes for action recognition purposes is a challenging task, as compression is to be achieved without losing the impression of the action. Insufficient or wrong keyframes can lead to confusion in the type of action. Proposed two-level key frame extraction algorithm uses an adaptive threshold technique to identify most dissimilar frames as keyframes. At the first level, global features based on intensity histogram and at second level local features computed from wavelet decomposition are used as similarity measures. A new performance parameter, Compression Ratio-Normalized Fidelity (CRNF) is introduced for evaluating the keyframe extraction algorithm. Average CRNF of 0.81 is achieved by the proposed algorithm, as compared 0.37 achieved by existing histogram-based method for action recognition dataset. CRNF value of 0.95 is achieved for Open Video project dataset, which is more than existing methods. Qualitative results further prove the effectiveness of the proposed algorithm.

International Journal of Intelligent Engineering and Systems

Two Level Key Frame Extraction for Action Recognition Using Content Based Adaptive Threshold

Vehicle safety is the most important strategy used to address the international and national road casualty reduction targets to achieve a safer road traffic system. Traffic accidents are mainly caused by human mistakes such as lack of concentration, misbehavior and distraction. The vision based lane detection system is a valuable way to protect single vehicle from accident. In this paper a lane detection method for detecting the lanes accurately is proposed. Also the canny algorithm is determined as the edge detection algorithm and Hough transform method is selected as the efficient way to detect lanes. To meet the real requirement, the region of interest (ROI) is defined to reduce noise properly for accurate results and to enhance the processing speed.

International Journal of Advanced Trends in Computer Science and Engineering

Straight and curve lane detection system for car safety using hough transform and sobel operator

Human action recognition has become vital aspect of video analytics. This study explores methods for the classification of human actions by extracting silhouette of object and then applying feature extraction. The method proposes to integrate HOG feature and PCA feature effectively to form a feature descriptor which is used further to train KNN classifier. HOG gives local shape-oriented variations of the object while PCA gives global information about frequently moving parts of human body. Experiments conducted on Weizmann and KTH datasets show results comparable with existing methods. © 2018, Springer Nature Singapore Pte Ltd.

Advances in Intelligent Systems and Computing

Silhouette-based human action recognition by embedding HOG and PCA features

In this paper, we propose a method for detecting moving objects in a video stream. Initially four existing methods, Basic Background Subtraction (BSS), Frame Differencing (FD), Adaptive Background Subtraction (ABS) and Background Subtraction Frame Differencing (BSFD) are evaluated for low resolution action recognition videos. Focus of this paper is on low resolution videos as most of the time CCTV recordings of the public areas are of low resolution. Recognizing an action performed by a human being in such a video is a challenging task. Videos recorded with various view angles are used for testing to evaluate view invariance property of algorithms. Concept of three frame differencing method is used along with adaptive background in an Adaptive Background subtraction and Frame Differencing (ABSFD) method proposed in this paper. A dynamic background image is used to detect the moving object. Background is updated with small constant value to accommodate minute changes in background. The proposed algorithm gives consistent results for videos recorded at different view angles. The experimental results show that this algorithm can detect moving objects successfully with precision above 95% in most of the test cases. © 2016 Institution of Engineering and Technology. All rights reserved.

IET Conference Publications

Evaluation of moving object detection techniques for low resolution videos

In this paper we propose a novel face detection approach. In general, lighting in images can have negative impact on the performance of face detection systems. Many current face detection systems are only reliable under controlled condition such as indoors under stable lighting conditions. So there is need to devise a solution to overcome illumination problems faced by face detection system. As one of the solution proposed approach firstly adopts a 'reference white' to realize light compensation; Then it realizes the detection of candidate face region using Ycbcr skin color model. The third, paper uses well known opening and closing morphological operations to exclude the impact of noise on color segmentation. Last Adaboost algorithm is improved and used to accurate face detection. The system uses java to achieve and is verified in self designed photo library, and is particularly suitable for illumination variations and complex background. © 2012 IEEE.