Perceptual Video Quality Estimator over a noisy wireless network
Incorporate the human perception to enhance the perceptual quality of the represented visual signals
Overcome the difficulties to integrate computational models with coding components
Improve the perceptual quality for constrained video coding whilst satisfying various constraints
Video Shot Detection
Organized and segmented video sequences into coherent units for video indexing
Implemented robust algorithms using cinematic techniques for video scenes
Extracted real time features and combined multiple features for comparison among successive frames
MPEG-2 Encoder and Decoder
Designed friendly interfaces, front panels, and included the support for different display sizes
Processed high fidelity encoders and Integrated Receiver Decoder for news and premium events
Deployed MPEG-2 video to an embedded ARM Processor based video board
Video Compression Algorithm Design and Implementation
Designed an efficient predictive coding algorithm for intra coding of video sequences
Designed a low-complexity entropy coding algorithm
Designed and implemented a video decoding system that optimizes the storage and computation costs
Load balancing and Fairness
Tracked project resources and easily managed them with automatic usage of resources
Illustrated instant views into the weekly and monthly allocation of all of the available resources
Identified resource shortages and excesses, and booked resources proportionately based on availability
Memory and Processor Scheduler
Maintained fairness, throughput, responsiveness, and scalability of CPU and memory resources
Implemented the best practices of CPU scheduling to get the most of a scheduling policy
Enhanced system throughput by adjusting specifications like CPU shares, reservations, and limits
Modified K-Means
Overcame the limitations and combined the centroids to improve the clustering step
Conveyed intuitions about the algorithm behavior useful for debugging and engineering new systems
Reduced the possibility of converging at local minima and estimated the correct number of clusters
Fargo-Moorhead Health Services
Viewed, in real time, reports of arrived, checked-in, remaining, and no-show patients
Communicated to the patients and doctors using social media by sending a reminder of the appointment
Facilitated and provided logistics to deliver quality solutions through web technologies
Memory and Processor Scheduler
Maintained fairness, throughput, responsiveness, and scalability of CPU and memory resources
Implemented the best practices of CPU scheduling to get the most of a scheduling policy
Neural Networks for Visual Recognition
Understood vision models and building new models or improving existing models for a vision task
Minimized the loss function with respect to the parameters of the score function
Mapped the pixel values to confidence scores and classifying test image by a single matrix multiplication