Parveen Lehana’s research while affiliated with University of Jammu and other places

Dermatoglyphics, the study of unique ridge patterns on fingertips, plays a crucial role in fingerprint-based identification. However, skin conditions such as psoriasis, eczema, and verruca vulgaris can distort these patterns. This distortion creates challenges for fingerprint matching systems in forensic science and biometric authentication. This paper proposes a Siamese network with a modified ResNet50 architecture and multihead attention mechanisms to improve fingerprint matching under such distortions. The Siamese network enables the system to compare fingerprint pairs effectively, even under distortions. The modified ResNet50 architecture captures intricate ridge patterns, while the multihead attention mechanisms focus on critical fingerprint regions. Overall, the proposed approach enhances the system’s ability to learn discriminative features. As a result, it can effectively differentiate between matched and unmatched fingerprint pairs, even in the presence of moderate to extreme noise. The proposed system was trained on the Sokoto coventry fingerprint dataset (SOCOFing) and a custom dataset. Experimental results demonstrate high accuracy, with the system achieving 99.47% accuracy under low noise conditions and 90% under moderate noise. However, performance declined to 55.56% under extreme distortions. A comparative analysis highlighted the superiority of the proposed system over widely used minutiae-based methods. The latter exhibited a significant drop in matching scores, falling below 15% for highly distorted fingerprints. These findings underscore the limitations of traditional fingerprint recognition techniques and highlight the effectiveness of proposed approach in handling dermatoglyphic distortions.

Face identification is amongst the most efficacious and extensive applications in biometrics involving extraction and locating facial features. With identification being monotonous task attributable to reliance on parameters like varied cameras, fluctuating backgrounds, and exposure to the environment in which an individual is present. Thermal imaging is endeavoring to resolve the accuracy issue of apparent imaging, such as lighting and brightness intensity, among all biometric variables. This paper presents a study of thermal imaging and effective methods involved in the feature extraction process for facial features with thermal imaging under the influence of varied noise. A novel face dataset is created TID comprising 27 thermal images and its corresponding visual band image using Fluke 480 Ti Pro camera. The study analyses detection efficiency of six feature extraction techniques in visible and thermal bands in facial features identification. Also, the influence of noise in the thermal band within the region of interest using feature points FIN, FOUT has been estimated. Throughout TID dataset, ORB extraction technique has been able to identify strongest inlier features FIN to a maximum extent with detection around the nose, eyes, and mouth. Further, results indicate feature detection in thermal images being invariant to effect of noise for detecting facial features.</p

Precise and timely detection of a crop’s nutrient requirement will play a crucial role in assuring optimum plant growth and crop yield. The present study introduces a reliable deep learning platform called “Deep Learning-Crop Platform” (DL-CRoP) for the identification of some commercially grown plants and their nutrient requirements using leaf, stem, and root images using a convolutional neural network (CNN). It extracts intrinsic feature patterns through hierarchical mapping and provides remarkable outcomes in identification tasks. The DL-CRoP platform is trained on the plant image dataset, namely, Jammu University-Botany Image Database (JU-BID), available at https://github.com/urfanbutt. The findings demonstrate implementation of DL-CRoP—cases A (uses shoot images) and B (uses leaf images) for species identification for Solanum lycopersicum (tomato), Vigna radiata (Vigna), and Zea mays (maize), and cases C (uses leaf images) and D (uses root images) for diagnosis of nitrogen deficiency in maize. The platform achieved a higher rate of accuracy at 80–20, 70–30, and 60–40 splits for all the case studies, compared with established algorithms such as random forest, K-nearest neighbor, support vector machine, AdaBoost, and naïve Bayes. It provides a higher accuracy rate in classification parameters like recall, precision, and F1 score for cases A (90.45%), B (100%), and C (93.21), while a medium-level accuracy of 68.54% for case D. To further improve the accuracy of the platform in case study C, the CNN was modified including a multi-head attention (MHA) block. It resulted in the enhancement of the accuracy of classifying the nitrogen deficiency above 95%. The platform could play an important role in evaluating the health status of crop plants along with a role in precise identification of species. It may be used as a better module for precision crop cultivation under limited nutrient conditions.

One of the most intriguing areas of computer vision is object detection. Due to variances in the environments, precise object recognition is a difficulty. The ability to track an item under occlusion scenarios is another challenging task. When an object has been identified, an essential role is to construct boundary boxes around it. An occluded environment hinders the identification process leading to the misclassification of diverse object types. With CNNs being the utmost and most efficient technique for object identification, this paper analyzes the performance of masked R-CNN for object identification in the presence of distortions like blur, noise, and contrast. The experiment has been conducted on benchmarked MS-COCO dataset containing multiple object classes. Results depicted the influence of distortions have indeed affected the identification competency of the algorithm with white noise affecting the identification capability to the maximum.

Convolutional neural networks (CNN) trained using deep learning (DL) have advanced dramatically in recent years. Researchers from a variety of fields have been motivated by the success of CNNs in computer vision to develop better CNN models for use in other visually-rich settings. Successes in image classification and research have been achieved in a wide variety of domains throughout the past year. Among the many popularized image classification techniques, the detection of plant leaf diseases has received extensive research. As a result of the nature of the procedure, image quality is often degraded and distortions are introduced during the capturing of the image. In this study, we look into how various CNN models are affected by distortions. Corn-maze leaf photos from the 4,188-image corn or maize leaf Dataset (split into four categories) are under consideration. To evaluate how well they handle noise and blur, researchers have deployed pre-trained deep CNN models like visual geometry group (VGG), InceptionV3, ResNet50, and EfficientNetB0. Classification accuracy and metrics like as recall and f1-score are used to evaluate CNN performance.

Face recognition (FR) is one of the most active research areas in the field of computer vision. Convolutional neural networks (CNNs) have been extensively used in this field due to their good efficiency. Thus, it is important to find the best CNN parameters for its best performance. Hyperparameter optimization is one of the various techniques for increasing the performance of CNN models. Since manual tuning of hyperparameters is a tedious and time-consuming task, population based metaheuristic techniques can be used for the automatic hyperparameter optimization of CNNs. Automatic tuning of parameters reduces manual efforts and improves the efficiency of the CNN model. In the proposed work, genetic algorithm (GA) based hyperparameter optimization of CNNs is applied for face recognition. GAs are used for the optimization of various hyperparameters like filter size as well as the number of filters and of hidden layers. For analysis, a benchmark dataset for FR with ninety subjects is used. The experimental results indicate that the proposed GA-CNN model generates an improved model accuracy in comparison with existing CNN models. In each iteration, the GA minimizes the objective function by selecting the best combination set of CNN hyperparameters. An improved accuracy of 94.5 % is obtained for FR.

The phenotyping of plant roots is a challenging task and poses a major lacuna in plant root research. Roots rhizospheric zone is affected by several environmental cues among which salinity, drought, heavy metal and soil pH are key players. Among biological factors, fungal, nematode and bacterial interactions with roots are vital for improving nutrient uptake efficiency in plants. The subterranean nature of a plant root and the limited number of approaches for root phenotyping offers a great challenge to the plant breeders to select a desirable root trait under different stress conditions. Identification of key root traits can provide a basic understanding for generating crop plants with enhanced ability to withstand various biotic or abiotic stresses. For instance, crops with improved soil exploration potential, phosphate uptake efficiency, water use efficiency and others. Laboratory methods such as hydroponics, rhizotron, rhizoslide and luminescence observatory for roots do not provide precise and desired root quantification attributes. Though 3D imaging by X-ray computed tomography (X-ray-CT) and magnetic resonance imaging techniques are complex, however, it provides the most applicable and practically relevant data for quantifying root system architecture traits. This review outlines the current developments in root studies including recent approaches viz. X-ray-CT, MRI, thermal infrared imaging and minirhizotron. Although root phenotyping is a laborious procedure, it offers multiple advantages by removing discrepancies and providing the actual practical significance of plant roots for breeding programs.