Gulraiz Khan

Road Sign Recognition (RSR) is extensively used in autonomous driving assistance systems and intelligent transportation. In real-world apps, road signs are often occluded or obstructed by natural and/or artificial objects. Detecting and classifying road signs becomes increasingly challenging in the presence of natural or artificial occlusions and obstructions. This study introduces two phased framework to address occluded road signs. In the first phase, based on ablation study an enhanced U-Net model is proposed, trained with composite loss function included structural similarity index, area of interest edge and attention based skip connections. This enhanced U-Net yielded improved Inpainting quality in comparison with baseline U-Net with SSIM improved to 0.817 from 0.797, MS-SSIM to 0.813 from 0.798. LPIPS to 0.257 from 0.291. Furthermore, computational cost reduced raw pixel wise PSNR to 21.98 dB from 22.40 reflecting overall performance in retaining structural details of processed image.In second phase, study explored baseline and inpainted variants of YOLOv5 and Faster R-CNN under variable occlusion thresholds. We tested all four models with incremental occlusion thresholds in road signs, ranging from 0% to 80%, and analyzed their respective impacts on learning in YOLOv5 and Faster R-CNN. The results show a significant improvement in detection when models are strengthened with inpainting techniques. While engaging Inpainting achieved with proposed enhanced U-Net, YoloV5 recall improved by 27.1 percentage points in the presence of 40% occlusion threshold. At 50% occlusion threshold, 30 points improvement is observed, 27.9 points at 60% occlusion level. .

Recent times have witnessed an exponential increase in multimedia specifically visual contents. Emotions are considered an essential part for extracting facial features, evaluating the expressions and as a result predicting the emotions of any person is a trending topic of the time. Based on still images and consecutive video frames, a methodology has been proposed to anticipate the emotions. Facial action coding system (FACS) standards are utilised in the development of an automated visual based emotion detection system worldwide. Employing FACS, the authors estimated facial muscle movement by computing 24 landmark points, 16 mutual distances between them and wrinkles caused due to changing expressions. Canny edge detection has been deployed to calculate the intensity of wrinkles. Geometric positions and optical flow are the key methods deployed in the implemented methodology. The methodology was evaluated on self‐generated, JAFFE dataset and EmotioNet.