Ishita Arora’s scientific contributions

Human Activity Recognition (HAR) has recently attracted the attention of researchers. Human behavior and human intention are driving the intensification of HAR research rapidly. This paper proposes a novel Motion History Mapping (MHI) and Orientation-based Convolutional Neural Network (CNN) framework for action recognition and classification using Machine Learning. The proposed method extracts oriented rectangular patches over the entire human body to represent the human pose in an action sequence. This distribution is represented by a spatially oriented histogram. The frames were trained with a 3D Convolution Neural Network model, thus saving time and increasing the Classification Correction Rate (CCR). The K-Nearest Neighbor (KNN) algorithm is used for the classification of human actions. The uniqueness of our model lies in the combination of Motion History Mapping approach with an Orientation-based 3D CNN, thereby enhancing precision. The proposed method is demonstrated to be effective using four widely used and challenging datasets. A comparison of the proposed method’s performance with current state-of-the-art methods finds that its Classification Correction Rate is higher than that of the existing methods. Our model’s CCRs are 92.91%, 98.88%, 87.97.% and 87.77% which are remarkably higher than the existing techniques for KTH, Weizmann, UT-Tower and YouTube datasets, respectively. Thus, our model significantly outperforms the existing models in the literature.

Human pose estimation and detection are critical for understanding human activities in videos and images. This paper presents a novel approach to meet the advanced demands of human–computer interactions and assisted living systems through enhanced human pose estimation and activity recognition. We introduce IMPos-DNet, an innovative technique that integrates multi-person pose estimation and activity recognition using a 3D Dual Convolution Neural Network (CNN) applied to multiview video datasets. Our approach combines top-down and bottom-up models to improve performance. The top-down network focuses on evaluating human joints for each individual, enhancing robustness against inaccurate bounding boxes, while the bottom-up network employs normalized heatmaps based on human detection, improving resilience to scale variation. By synergizing the 3D poses estimated by both networks, IMPos-DNet produces precise final 3D poses. Our research objectives include advancing the accuracy and efficiency of pose estimation and activity recognition, as well as addressing the scarcity of 3D ground-truth data. To this end, we employ a semi-supervised method, broadening the model’s applicability. Comprehensive experiments on three publicly available datasets—Human3.6 M, MuPoTs-3D, and MPI-INF-3DHP—demonstrate the model’s superior accuracy and efficiency. Evaluation results confirm the effectiveness of IMPos-DNet’s individual components, highlighting its potential for reliable human pose estimation and activity recognition.

Video saliency detection has recently been widely used because of its ability to distinguish significant regions of interest. It has several applications, such as video segmentation, abnormal activity detection, video summarization, etc. This research paper develops a novel technique for video saliency detection known as Spatiotemporal Recurrent Fully Convolutional Network Model (SRFCNM). This model uses recurrent convolutional layers to represent spatial and temporal features of superpixels for element uniqueness. The model is trained in two phases; initially, we pre-train the model on the segmented data sets and then fine-tune it for saliency detection, which allows the network to learn salient objects more accurately. The uniqueness of integrating saliency maps with recurrent convolutional layers and spatiotemporal characteristics facilitates the robust representation of salient objects to capture the relevant features. The SRFCNM model is extensively estimated on the challenging datasets viz. SegTrackV2, FBMS and DAVIS. Our model is compared with the existing Deep Learning and Convolutional Neural Network algorithms. This research demonstrates that SRFCNM outperforms the state-of-the-art saliency models considerably over the three public datasets regarding accuracy recall and mean absolute error (MAE). The proposed SRFCNM model produces the lowest MAE values, 3.2%, 3.5%, and 7.5%, for SegTrackV2, DAVIS, and FBMS datasets, respectively, with hand-crafted color features, compared with the existing models.