NUST Institutions Library Catalogue Search for 'an:"119537"'

Urdu Digital Text Optical Character Recognition /

By Ahmed, Mustafa. . 46p. ; 30cm..

Place Hold on Urdu Digital Text Optical Character Recognition /

Deep Learning Methods for Disease Identification of Cotton Plants /

By Fasihi, Sajeel . . 79p. ; 30cm..

Place Hold on Deep Learning Methods for Disease Identification of Cotton Plants /

Traffic Signal Control using Reinforcement Learning /

By Umer Jamil, Qazi . . 90p. ; 30cm..

Place Hold on Traffic Signal Control using Reinforcement Learning /

Classification of Live Video Stream from Pakistani News Channels (Urdu) using Deep Learning Latest Techniques /

By Afzal, Muhammad . . 134p. , In our contemporary era, information is of prime importance and its dominant use by social media and TV channels for making public opinion and cultural influence is quite evident. Videos form the major portion of media and contain more elaborate information than a single image. Today, videos are piling up in millions every day and their segregation, classification and analysis are upheaval tasks. Live TV video stream contains voice, metadata and image frames full of multiple information including written scripts etc. which can contribute to video classification. But utilization of each type of data we need to do a separate study. However, we have focused on classification of video stream using deep learning (DL) neural networks which are well established solutions for images and small videos classification and gesture recognition. In our study, we have suggested a mechanism for classification of big or live video streams obtained from Pakistani TV News Channels into 5 classes (Advertisement, News, Talk Show, Sports & Entertainment Program) using supervised DL pretrained neural networks. Due to non-availability of authentic dataset on this subject, we have created a customized data of videos recorded (approximately 335 hours videos) from various sources like different TV channels‘ websites and YouTube. Videos were processed to extract image frames to prepare a trainable dataset. For our experimentation, we have mainly used pretrained ResNet variants (ResNet18, ResNet34, ResNet50, ResNet101 & ResNet152) on ImageNet dataset and few other models like AlexNet, ConvNeXt_Tiny, DenseNet121, SqueezeNet and VGG11 for comparison purposes. Then modified the last classification layer of the network as per number of target classes and finetuned all weights of neural network on the subject dataset. We carried out various experiments on these neural networks and achieved quite encouraging results having accuracies ranging from 95% to 99%. For testing of videos on trained models, dynamic averaging time domain window was applied to diminish the jitters in the output results. This can be useful in many other applications as well including social media & advertisements analysis, classification of small videos, industrial and business automation etc. 30cm.

Place Hold on Classification of Live Video Stream from Pakistani News Channels (Urdu) using Deep Learning Latest Techniques /

Pose-Based Seamless Video Stitching for Real World Applications /

By Hassan, Salman . . 70p. , Combining videos of humans performing different gestures in a smooth way can potentially have many uses across a wide range of fields. These include entertainment, virtual reality, robotics, education, & communication. The goal of this research work is set in this context. This research focuses on developing a system that takes individual videos of humans performing motion gestures, and stitches them in a way that minimizes spatial discontinuities between upper torso joints, thus joining two or more human gestures into one seamless continuous motion. It begins by investigating & comparing current frameworks used to stitch individual human motion gestures and investigates the theoretical and mathematical approaches behind them, proceeding in a step-by-step way. First, it collects sign videos for most commonly used English sentences of lengths 2-8. Then, it preprocesses these videos to convert them into a standardized form. Following that, it extracts landmarks to prune unnecessary parts of videos. It then calculates human joint coordinates using pose estimation. After that it calculates link vectors and human shoulder, and elbow angles using linear algebra. Following that, the system interpolates joint coordinates at transitions between signs and uses them to calculate interpolated joint angles. Concurrently, actual joint coordinates are used to calculate actual joint angles, which are then used to calculate wrist poses using forward kinematics. These wrist poses are compared with the same obtained from feeding interpolated joint angles to forward kinematic models. An ablation study was then conducted that measured mean errors across different combinations of spline degree, percentage of knots, & length of sentences. LSQ Univariate Spline with degree 4, knots percentage of 90%, and sentence length of 4 produced least mean error. Transition errors (errors between sign transitions were also calculated & recorded for each of 100 sentences. In this way, smoothness of different interpolating functions was quantified 30cm.

Place Hold on Pose-Based Seamless Video Stitching for Real World Applications /

Stitch Multiple Images for Generating Quality Panorama /

By Riaz, Sibgha . . 53p. , Stitching multiple images for achieving the 360 view of any environment is a challenging task. Traditionally, the whole process of image stitching is based on distinctive features that are very helpful for estimating the other parameters of the whole algorithm. As different images require different suitable parameters or weights for achieving the best results and we need to predict those suitable parameters for each case independently. In our proposed model first small neural network based techniques are implemented that are just used for estimating the quality panorama hyper parameters and then we apply the whole stitching algorithm on sample images by using those predicted parameters. Therefore, due to lack of labeled data we are unable to train any supervised model for those hyper parameter selection that’s why we build an unsupervised technique that makes decisions based on just extracted features quality, confidence and count of inliers etc. By estimating the good parameters we are able to stitch a quality panorama that doesn't have any ghosting artifacts, blending discontinuities, seamless and alignment errors as well. We evaluate the performance of our proposed model on three datasets and analyze performance in both perspective quality and computational time and conclude that our model outperforms with other state of the art stitching algorithms in both perspectives. 30cm.

Place Hold on Stitch Multiple Images for Generating Quality Panorama /

Glossing of American Sign Language (ASL) /

By Thaheem, Muhammad Arslan . . 49p. , American Sign Language (ASL) is a sign language used in America with slight modifications across different regions. Over the years it has developed and included a lot of new signs in it. In order for the deaf community to take notes and communicate with common people, ASL glossing is done which is an organized sentence structure of ASL. The goal of this research is to make a rule-based engine that can convert English sentences into ASL Gloss. The research included three phases. Firstly, we collected the English to ASL sentences from different resources including books, websites, etc. Secondly, we made rule-based engine to standardize the format of glossing. Since there are no proper set of rules written till now for ASL, we extracted common rules from sentences collected from different sources and continued our research on the basis of that. Thirdly, the engine was checked using different English sentences. We were able to achieve a BLEU score of 20.85 on the test set. 30cm.

Place Hold on Glossing of American Sign Language (ASL) /

Design of a Non-linear Controller for a Multi-Rotor UAV /

By ZAHID , MUHAMMAD OMER . . 143p. ; 30cm..

Place Hold on Design of a Non-linear Controller for a Multi-Rotor UAV /

Effect of Image Resolution on The Classification of Cells in Peripheral Blood Smear /

By Ali, Adnan . . 55p. 30cm.

Place Hold on Effect of Image Resolution on The Classification of Cells in Peripheral Blood Smear /

Optimal Pose Estimation of Robotic Manipulator For Pick And Place Application /

By Farrukh, Uzair . . 46p. , This thesis examines how to use robotic manipulator to pick specific object. In order to attain above objective planning of manipulator as well as grasping point calculations are studied. The process of calculating grasping pose with respect to target object,the grasping point essential for successful grasping of object is called grip synthesis.Both analytical and empirical approaches are used to calculate grasping points. While empirical uses human like strategies, analytical approaches are mainly dependent on geometric, kinematic, and/or dynamic formulations. The main goal of the thesis is to minimize joint accelerations. Constrained optimization techniques like genetic algorithm and path search algorithm are used to find grasping point and minimize joint accelerations to ensure successful grasping of object.A novel hybrid approach is also proposed which used both genetic algorithm and path search algorithm to find its fitness value. The first contribution is finding optimal grasping points of regular shaped object by calculating centroid of object and are near to centroid along minor axis. The second contribution is to minimize joint accelerations by using optimization techniques of genetic algorithm and path search algorithm in which global minima’s are found and optimized trajectory is generated which helps in successful grasps with minimum joints accelerations. Third is the introduction of novel hybrid approach which uses both genetic algorithm and pattern search algorithm to find its global minima which is more robust for changing search space. 30cm.

Place Hold on Optimal Pose Estimation of Robotic Manipulator For Pick And Place Application /