Human Action Recognition (HAR) is a branch of computer vision that deals with the identification human actions at various levels including low level, action and interaction level. Previously, number HAR algorithms have been proposed based on handcrafted methods for recognition. However, techniques are inefficient in case recognizing level as they involve complex scenarios. Meanwhile, traditional deep learning-based approaches take entire image an input later extract volumes features, which...

Abstract Advancements in facial manipulation technology have resulted highly realistic and indistinguishable face expression swap videos. However, this has also raised concerns regarding the security risks associated with deepfakes. In field of multimedia forensics, detection precise localization image forgery become essential tasks. Current deepfake detectors perform well high-quality faces within specific datasets, but often struggle to maintain their performance when evaluated across...

Human action recognition (HAR) is one of the most active research topics in field computer vision. Even though this area well-researched, HAR algorithms such as 3D Convolution Neural Networks (CNN), Two-stream Networks, and CNN-LSTM (Long Short-Term Memory) suffer from highly complex models. These involve a huge number weights adjustments during training phase, consequence, require high-end configuration machines for real-time applications. Therefore, paper presents an extraneous frame...

In this paper, a new computation for gray level co-occurrence matrix (GLCM) is proposed. The aim to reduce the burden of original GLCM computation. proposed will be based on Haar wavelet transform. transform chosen because resulting bands are strongly correlated with orientation elements in second reason total pixel entries always minimum. Thus, can reduced. tested classification performance Brodatz texture images. Although achieve at least similar computation, gives slightly better compare

Research in human activity recognition (HAR) has seen tremendous growth and continuously receiving attention from both the Computer Vision Image Processing communities. Due to existence of numerous publications this field, undoubtedly, there have been a number review papers on subject that categorise these techniques. Many recent works started tackle more challenging problems proposed techniques are addressing realistic real‐world scenarios. Conspicuously, an updated survey covers methods is...

Neuroimaging plays an important role in the diagnosis brain lesions such as tumors, strokes and infections. Within this context, magnetic resonance diffusion-weighted imaging (DWI) is clinically recommended differential of several by providing detailed information regarding lesion based on diffusion water molecules. Conventionally, performed visually professional neuroradiologists during a highly subjective, time-consuming process. In response, computer-aided detection/diagnosis (CAD) has...

Skin detection is a key aspect of many computer vision applications including face detection, person identification, illicit content and other related applications. In this paper, skin method proposed combining two color spaces HSV (Hue, Saturation, Value) YCgCr (luminance, chrominance in green, red). The S, Cg Cr components are used to form hybrid SCgCr space. results show that, the can respond well different tones with less sensitivity skin-like background pixels. It also shown that higher...

In this paper, an adaptive thresholding technique based on gray level co-occurrence matrix (GLCM) is presented to handle images with fuzzy boundaries. As GLCM contains information the distribution of transition frequency and edge information, it very useful for computation threshold value. Here algorithm designed have flexibility definition so that can object’s By manipulating in GLCM, a statistical feature derived act as value image segmentation process. The proposed method tested starfruit...

Malaysian car plates in general appear different character styles, types (either single or double row), sizes, spacing and counts. Such variations cause even detecting localizing these a difficult problem. The problem of localization is aggravated further during night time due to poor illumination. In this paper, we introduce the idea edge-geometrical features plates. edge part obtained from use Difference Gaussian operation followed by Sobel vertical mask. Prior that, gamma correction...

Crowd counting considers one of the most significant and challenging issues in computer vision deep learning communities, whose applications are being utilized for various tasks. While this issue is well studied, it remains an open challenge to manage perspective distortions scale variations. How these problems resolved has a huge impact on predicting high-quality crowd density map. In study, hybrid modified neural network (U-ASD Net), based U-Net adaptive scenario discovery (ASD), proposed...

Visual motion segmentation (VMS) is an important and key part of many intelligent crowd systems. It can be used to figure out the flow behavior through a spot unusual life-threatening incidents like stampedes crashes, which pose serious risk public safety have resulted in numerous fatalities over past few decades. Trajectory clustering has become one most popular methods VMS. However, complex data, such as large number samples parameters, makes it difficult for trajectory work well with...

This paper presents a novel face detection system based on feature-based chrominance colour information from an image containing one in indoor environment with non-uniform background. The algorithm is the adapted chain code (ACC), eye and modified golden ratio (MGR). ACC proposed to obtain boundary. MGR attempts extract part of that includes eyes, eyebrows, nose mouth, detected eyes' positions. Experimental results show able detect near frontal high accuracy. database consists faces without...

This paper presents an efficient technique for real-time tracking of a single moving object in terrestrial scenes using stationary camera. The algorithm is based on the linear prediction (LP) solved by maximum entropy method (MEM). It attempts to predict centroid next frame, several past measurements. Using second order method, proposed able accurately track object. shown analytically that recursive predictor-corrector yield high accuracy performance and superior Kalman filter, possibly...

In this paper, an adaptive thresholding technique based on gray level co-occurrence matrix (GLCM) is presented to handle images with fuzzy boundaries. As GLCM contains information the distribution of transition frequency and edge information, it very useful for computation threshold value. Here algorithm designed have flexibility definition so that can object's By manipulating in GLCM, a statistical feature derived act as value image segmentation process. The proposed method tested starfruit...