Yawar Rehman's research while affiliated with NED University of Engineering and Technology, Karachi and other places

... Therefore, the algorithms need to handle the occlusion problem and incorporate an attention mechanism. Some computer vision algorithms [14][15][16][17] employ CNNs to extract features from images, utilize data augmentation methods to enrich the features of small targets, and address the resolution issue. However, these algorithms struggle to effectively leverage local contextual features in the presence of significant target deformations. ...

... A group of researchers have evaluated indoor mapping outcomes using the trolley-based platform equipped with onboard sensors and computing capabilities [12]. Besides the trolley-based systems, to perform scanning tasks on uneven surfaces such as floor stairs or rough outdoor terrain, many research works have combined several Hokuyo 2D laser scanners with additional dead reckoning sensors on backpack systems to create 2D/3D models of the environment [13]. Similar scanning techniques have been adopted on car-based mobile mapping systems for outdoor forestry surveys by integrating the Hokuyo scanner with Real Time Kinematic Global Navigation Satellite System (RTK-GNSS) module [14]. ...

... Structure from motion (SfM) creates a 3D model by merging a sequence of 2D images captured by a camera sensor from different locations and estimating the relative camera positions and orientations [24]. Creating a 3D scene using CRP involves camera calibration, sequential picture capture, feature matching, spatial relationships, dense alignment, surface reconstruction, and texture matching [9,20]. ...

... Spectrum scarcity issue in vehicular communication can be addressed by incorporating cognitive radio spectrum sensing model. For maximizing throughput in VANET communication, optimistic spectrum sensing model is introduced in [4] that senses an optimal vehicle and machine learning approach is used for clustering the nodes to reduce communication burden. Combining VANET and internet of things (IoT) suggests internet of vehicle (IoV) and trust management is one of the main challenge in vehicular communication it plays vital role for providing security which needs responsive and adaptive methods to confirm security. ...

... Biometric systems based on the combination of two or more characteristics, referred to as multimodal systems, have several advantages compared to their unimodal counterparts as they allow improved recognition rate, universality, and the authentication of users for which one of the single biometric characteristic cannot be detected [1][2][3]. In particular, multimodal systems based on a single sensor are arousing interest because they permit to achieve cost-effectiveness and improved acceptability from users [4]. ...

... The subset images were enhanced by applying sharpness, brightness, contrast, gamma correction, and saturation filter values to 0.70, 0.80, 0.90, 1.10, and 1.20, respectively ( Table 1). The sharpness enhancement improves the object edge in an image and renders them more sensitive to detection [53]. Image sharpness was increased by setting the filter value to 10 (Table 1). ...

... When the 3D coordinate information is acquired by computer vision, the distortion coefficients and camera parameters e.g. the focal length is needed to be calculated beforehand. However, in [17], the camera calibration is implied by developing the 3D point of scanning object, using the structure from motion technique without using any specific model of the camera. Similarly, checkerboard corners are intelligently used to estimate the pose of a camera [18]. ...

... Then, he combined HOG features and SVM extraction to achieve an accuracy of 99.15% in GTSRB. Yawar Rehman [21] learned features by gathering semantic information, which reduced the use of sliding window in traditional target detection and relied on color changes to find areas with a higher possibility of traffic signs, achieving an accuracy of 100% on the GTSRB dataset. The deep learning method requires prior training of features from a large amount of sample data. ...

... The presence of obstacles in front of a sign might make it difficult to recognize because of the reduced visibility. This is a very common problem that can be caused by pedestrians, trees, other vehicles etc., [20]. Previous systems present relevant solutions for this issue, but have a very limited scope [19]. ...

... The GFA in this paper utilizes guided filter [15] to optimize the soft matting [20] so as to reduce time complexity, but still can not achieve real-time performance. Owing to the rise of convolutional neural networks, a plenty of researchers devote themselves to designing lightweight enhancement networks [21]- [25] to take the place of traditional algorithms so as to achieve a reduction in processing time while retaining effectiveness. For the consideration that for some non-severely defocused images, sacrificing some enhancement effectivity to ensure real-time performance is feasible, therefore the lightweight MDC-Net is designed as a supplement for the poor real-time performance of GFA. ...