We propose a novel method that combines the strengths of two popular class activation mapping techniques, GradCAM++ and ScoreCAM, to improve interpretability localization convolutional neural networks (CNNs). Our proposed method, called "Grad++-ScoreCAM", first utilizes algorithm generate coarse heatmap an input image, highlighting regions importance for particular class. Then, we employ ScoreCAM refine by incorporating information from intermediate layers network. By combining these can...

Noise or artifacts in an image, such as shadow artifacts, deteriorate the performance of state-of-the-art models for segmentation image. In this study, a novel saliency-based region detection and image (SRIS) model is proposed to overcome problem existence noise intensity inhomogeneity. Herein, adaptive level-set evolution protocol based on internal external functions designed eliminate initialization sensitivity, thereby making SRIS robust contour initialization. energy function, weight...

Active contour models have achieved prominent success in the area of image segmentation, allowing complex objects to be segmented from background for further analysis. Existing can divided into region-based active and edge-based models. However, both use direct data achieve segmentation face many challenging problems terms initial position, noise sensitivity, local minima inefficiency owing in-homogeneity intensities. The saliency map an changes representation, making it more visual...

Fashion image analysis has attracted significant research attention owing to the availability of large-scale fashion datasets with rich annotations. However, existing deep learning models for often have high computational requirements. In this study, we propose a new model suitable low-power devices. The proposed network is one-stage detector that rapidly detects multiple cloths and landmarks in images. designed as modification EfficientDet originally by Google Brain. simultaneously trains...

Surveillance videos are crucial for crime prevention and public safety, yet the challenge of defining abnormal events hinders their effectiveness, limiting applicability supervised methods. This paper introduces an unsupervised end-to-end architecture video anomaly detection that applies spatial temporal features to identify anomalies in surveillance footage. The model employs a three-dimensional (3D) convolutional autoencoder, with encoder-decoder structure learns spatiotemporal...

The most fatal and frequent cancer amongst women is breast cancer. Mammography provides timely detection of lumps masses in tissue, but effective diagnosis requires accurately identifying malignant tumor boundaries, which remains challenging, particularly for images with inhomogeneous regions. Therefore, we propose an active contour method based on a reformed combined local global fitted function to address segmentation. This strengthened by proposed average energy driving capture obscure...

The segmentation of images under biased conditions such as low contrast, high-intensity inhomogeneity, and noise is a challenge for any image model. ideal model must be capable segmenting with maximum accuracy minimum false-positive rate conditions. In this paper, we propose region-based active contour (ACM), called global signed pressure K-means clustering based on local correntropy the exponential family (GSLCE), to address challenges An adaptive weighted function formulated differences...

Level set models are suitable for processing topological changes in different regions of images while performing segmentation. Active contour require an empirical setting initial parameters, which is tedious the end-user. This study proposes incremental level model with automatic initialization contours based on local and global fitting energies that enable it to capture image containing intensity corruption or other light artifacts. The region-based area length terms use signed pressure...

Image segmentation is a tedious task that suffers from constraints, such as blurred or weak edges and intensity inhomogeneity. Active contour models, including edge-based region-based methods, are extensively used for image segmentation. Each of these methods has its pros cons affect image-segmentation accuracy CPU processing time. This study combines local global fitting energies uses statistical information to drag contours toward object boundaries, thus overcoming The bias field, the...

Image inhomogeneity often occurs in real-world images and may present considerable difficulties during image segmentation. Therefore, this paper presents a new approach for the segmentation of inhomogeneous images. The proposed hybrid active contour model is formulated by combining statistical information both local global region-based energy fitting models. inclusion assists extracting intensity regions, whereas curve evolution over homogeneous regions accelerated including method. Both...

Class activation maps (CAMs) are powerful tools for better understanding what convolutional neural networks learn and the reliability of their learning capability within relevant contexts.Highly inspired by Grad-CAM Score-CAM, this manuscript proposes a novel method called Increment-CAM, which overcomes limitations both methods enhances localization heatmaps, suppressing false activations to provide discriminative visualizations.This paper offers three-phase approach, generating through in...

Segmentation accuracy is an important criterion for evaluating the performance of segmentation techniques used to extract objects interest from images, such as active contour model. However, can be affected by image artifacts intensity inhomogeneity, which makes it difficult with inhomogeneous intensities. To address this issue, paper proposes a hybrid region-based model images. The proposed energy functional combines local and global functions; incorporated weight function parameterized...

Image generation is an important area of artificial intelligence that involves creating new images from existing datasets. It learning the distribution target randomly generated vectors. Like other deep models, image model requires a vast refined data set to produce high-quality results. When there little data, problem diversity and quality are compromised. In this paper, we propose generative applies PCA generator least square error adversarial network that, in turn, generates even with...

Region based active contours algorithms are extensively utilised for image segmentation irrespective of unavailability the densely annotated large data sets as required in case fully supervised deep learning models. However, previous models have certain limitations including false appearances when there is in-homogeneity image. In our model we combine local and global information level set function, proposing a hybrid energy function which helps efficiently evolve on may assess significance...

Object detection using vision transformers (ViTs) has recently garnered considerable research interest. Vision Transformers execute image classification through a multi-head attention-based MLP head and post-image segmentation into patches. However, conventional models prioritize object over predicting bounding boxes crucial for precise detection. To address this gap, two-stage detector been devised based on Transformers, which initially extracts feature maps via pre-trained CNN model. In...

Image segmentation is a crucial stage of image analysis systems because it detects and extracts regions interest for further processing, such as recognition the description. However, segmenting images not always easy accuracy depends significantly on characteristics, color, texture, intensity. inhomogeneity profoundly degrades performance models. This article contributes to literature by presenting hybrid Active Contour Model (ACM) based Signed Pressure Force (SPF) function parameterized...

Generative Adversarial Networks (GAN) is a research-based on deep learning technology that synthetically generates, combines, and transforms images similar to the original images. The main focus of GAN existing work has been improve quality generated generate high-resolution by changing training scheme or devising more complex models. However, these models require large amount data are not suitable for with small data. To address challenges, this paper aims stability dataset proposing novel...

There are significant and far-reaching implications for agriculture due to the extraordinary changes in global climate. Temperature variations affect structure of soil, precipitation patterns harsh weather lead soil erosion other negative effects. The agricultural sector faces various obstacles, ranging from fluctuations crop yields under climate stress difficulties managing water changing conditions. Integrating research, presenting local case studies, considering potential future paths may...

This study investigates the impact of, financial technology (fintech) integration, on efficiency, cost reduction, and, sustainability of energy markets, with a, particular focus on, Pakistan. Employing, Technology-Organization-Environment, (TOE) framework, research, explores how fintech, adoption, investment in, regulatory support, technological, infrastructure, and market transparency, influence key market, outcomes. A structured questionnaire, was distributed to professionals, stakeholders...

In the field of Artificial Intelligence, a large and densely annotated dataset is required for training making it time resource-expensive task. this paper, we propose an image generation network model that keeps examples at minimal level. The proposed gives additional feature maps to input value (latent space) DCGAN model, which adversarial using convolutional neural network. To solve problem cannot generate clear images in case lack data, one map was added latent space. extracted from 2,000...

Clothing detection and landmark are important techniques in fashion image analysis. The availability of large annotated datasets has made analysis a hot research topic. This paper proposes single-stage detector that performs bounding box detection, can also predict end-to-end clothing category classification. parallel processing provides improved time efficiency than the later technique regional proposals first then prediction module. proposed network is designed with revision EfficientDet...